1. 1. General Data Considerations
1.1. Data availability and criteria
1.1.1. *estimation of outgo to be as accurate as possible *Statistics used need to be: -Relevant - similar class of lives and policy conditions -Credible-large volume for precise estimates with little sampling error -Available *consider cost of collection *Format of data *apply judgement to compile statistics from multiple sources Q12.1 - useful
1.2. Making adjustments for credibility and relevance
1.2.1. *adjusments maybe requied, may have to apply non precise methodology and broad based ratios used initially *More accurate allowance when experience builds up
1.3. Projection to midpoint of insurance usage
1.3.1. Claim incidence rates
1.3.1.1. Incidents vary by age and calendar year *say estimate claim incidents rate ix(0), 0 is 2004. - probability of claim over period, for lives aged x *Launch new prod in 2005. Need to set rates assuming lives started policies half way in each year: 1/2[ix(2)+ix(3)] etc
1.3.2. Claim termination rates
1.3.2.1. Vary by duration from commencement of claim. Experience imrovement or deterioration over time may be assumed
1.3.3. Average claim size
1.3.3.1. PMI - function of inflation, current protocols, hospital charging structures, available of state treatment and care. Only look one year at a time, as prem reviewed. *If considerable future uncertainty, contracts allow for some form of review, else, cost of reserves can make premium too high to sell -margins in assumptions -cost of capital would be too high,. if not allowed for in price of contract, product not worth selling. *Smaller term, smaller margins.
1.3.4. IP and LTCI claim inception rates should be analyzed and projected CI separately by causes of claim PMI incidence and average claim costs analysed and projected rates projected to mid points of insurance usage
1.3.5. Group business
1.3.6. *data base will be historic. need to gather information on market, trends, and developments to project forward. *should incorporate policy term and length of time current terms will be offered *Assumptions need to stay valid for average of: [expected shelf life of proposed premium rates] +[ expected duration of policy to termination or to next review]
2. 2. Own data
2.1. Relevance
2.1.1. *May elements influencing amount to pay out so may not be reflected in external sources: -Underwriting approach -Policy conditions -Claims management -Distn channels -Target market own info best for prediciton, cost reliability, and format shouldn't be an issue
2.2. Credibility
2.2.1. *volume insuffice, credibility not there. *case when have to split into particular risk cells *May have to look external, but might not be relevant, there adjustments required. *spot trends externally, and understand relevance to own company. *can use statistical fitting methods, and also allow for further future changes *Own data has key benefit of relevance.
3. 3. Population data
3.1. Introduction
3.1.1. *govt produce periodic analysis of healthcare experience *sophisitication levels vary. *important for actuary, premium assessment, some risk cells, only source for which a first estimate can be gained. Drawbacks: -Accuracy and reliability maybe questionable, especially where subjective definitions, scope for double counting: CI cover heart attach and coronary artery bypass, PH only paid out once, but represented twice in population records -may not be available electronically or appropriate format -out of date before even published
3.2. Relevance
3.2.1. Drawbacks cont: -National experience may not be relevant to subset of lives that wil be accepted for insurance - may be improved if population analysis split by socio economic groups, regions, ages. big variation in morb by profession etc. -insured should be healthier then population as underwritten. But if not compulsary insurance, people choosing are arguably doing so as they expect to gain from it. Therefore, improvement in experience vs population might not be so marked. Evidence of Anti selection and moral hazard in territories. No universal rule to apply, adjustment to pop data will depend on specific details -Circumstances of onset and continuation of a disability are unlikely to coincide with claim definition. -underlying problem is we are not dealing with insured lives -different class of lives, the degree of sickness that triggers a claim different from pop stats
3.3. Credibility
3.3.1. Large volume, usually free
4. 4. Reinsurers data
4.1. Introduction
4.1.1. Offered as a part of a range of services *Reinsurer keen rthat statistics proposed are as accurate as possible, as profits under treaty reflected. *Reinsurers data will be used to help set insurers. Reins prem maybe based on these premiums. *Maybe explicit charge or provided on basis of accepting a share of profits from business written.
4.2. Relevance
4.2.1. -Assess divergence from own experience and make adjustments -Specifically to own underwriting and claims management
4.3. Credibility
4.3.1. Reinsurers can draw on involvement with other companies in the market, excellent position to assist company to launch new products. Breadth of exposure allows for direct pinpointing data most relevant to an insurer. Reinsurer won't be very credible for new products, as will grow as market does.
5. 5. Marketing data
5.1. Insured lives data
5.1.1. *data pooling - claims and policy statistics so industry wide information, more credible then individual access - CMI *Complied and published for all *Standard morbidity tables produced from here. advantages: -insured experience -sufficient volume for statistical credibility -reflects local companies and complied by experts. Disad: -time to publish final tables, reducing applicability -Market average hence, strictly not relevant -Strictness of underwriting and claims management might be reflected in similar ratios used for life insurance standard tables. if sell life as well as health, adjustment applied to mortality might be similar to morbidity.. -differences in policy conditions from the average may produce more straightforward ratios if the cost of risk has been well analysed - page 14 - recheck -Market data may be available for long term healthcare business, but short term very limited - PMI
5.2. Returns to insurance supervisor
5.2.1. detailed information on balance sheet, annual revenue, policies inforce, reserves, benefits, premiums by class, ons and offs during a year. -Accuracy not questioned, but extractable detail for estimation of future risk in individual cells is questionable. -Allows for high level check on premiums produced from other sources - useful when launching healthcare product whereby competitor rates are considered in context of solvency levels, profitability, and market positioning.
5.3. Relevance
5.3.1. *National relevance to company concerned *care required for policy conditions differing, underwriting and claims control can be materially different from market average.
5.4. Credibility
5.4.1. More credibile then own data, but not as much as population statistics.
6. 6. Other sources
6.1. Overseas
6.1.1. Relevance
6.1.1.1. Likely to be less relevant than equivalent home data *even with insured life data, care required for differing cultures, state health care, marketing practices, legislation, policy conditions.
6.1.2. Credibility
6.1.2.1. Maybe available, adjustments
6.1.3. May chose broad based population ratios to assess claims experience locally to that inherent in database available. *Estimate own by: {oversea indust morb rates}*{home pop morb rates}/{oversea pop morb rates}
6.2. Actuarial consultants
6.2.1. Access to data from national and international sources - like reinsurers Likely not to share in the risk directly, but will require reward in helping compile premiums
6.3. Rate table software
6.3.1. Software comparing market premium rates for similar products for a given class. Useful for gathering initial data or checking reasonableness and competitiveness of office premiums once calculated.
6.4. Trade magazines
6.4.1. National and global source of statistics in rsk projection process. lack detailed breakdown, but info for reasonableness check.