Forecasting and Modeling

Question 1

Uses of health insurance financial models (2)

Answer 1

1. Pricing – financial and sales models are used to determine premiums
2. Reserve calculations and reserve basis evaluation – some reserves (like gross premium reserves) are calculated by forecasting models
3. Monitoring of results – to validate assumptions, to warn of deviations from expected values, and for resource planning
4. Solvency testing – may indicate a need for gross premium reserves
5. Financial forecasting – corporation forecast results for various reasons
6. Actuarial appraisals – these are studies of the value of a block of business, typically used when transferring ownership

Question 2

Essential characteristics of a good model (2)

Answer 2

1. Reliable accuracy – a model must be good at predicting the future. It must also be robust
2. Suitability for use – the model should produce the results it is design for, without adding unnecessary complications
3. Appropriate precision – this relates to how many decimal places should be kept in the values
4. Sensibility – the model should reflect a logical construction of what is being modeled. It should also be theoretically sound
5. Effectively communicated – this includes communicating everything necessary to understand and use the model’s results

Question 3

Steps in building a forecast model (3)

Answer 3

1. Choosing the basic structure of the model
   a. Tools used include spreadsheets, database models, and sequential program
   b. Model types include asset share models, reserve development models, and agent-based models
2. Choosing the information to be carried – the information needed will depend on the purpose of the model
3. Choosing assumptions and building a prototype projection
   a. Starting values and assumptions must be built into the model
   b. A prototype cell is defined, and then projected to the end of the forecast period
4. Extending the prototype – after the prototype cell is built, the model must be extended to other cells which represent the different subsets of the business being modeled
5. Validating the model (see separate list)
6. Documenting the model – this allows the model to be evaluated by other professionals, and makes it easier to make modifications
7. Designing output and communicating results – the model output can be useless unless it is put into the context of the question being asked

Question 4

Methods for validating forecast models (6)

Answer 4

1. Starting values are compared directly to the actual values for that year
2. Year to year changes in the model are compared to actual past historical results
3. Model results are checked for reasonableness by people familiar with the business
4. Stress testing – analyze how the modeled results behave when some of the underlying assumptions are changes (includes sensitivity testing)

Question 5

Assumptions needed for forecasting (6)

Answer 5

1. Lapse – lapse rates vary widely by product, duration, company and member or policy characteristics. They are generally highest in the first year, then decrease thereafter
2. Mortality – some models treat mortality as a separate decrement, but most models combine mortality and lapses (mortality is a minor assumption for health insurance)
3. Claim costs – it’s best to use actual experience when possible. Trend assumptions are needed for determining future claim costs
4. Expenses – usually expressed on a per unit bases (per policy, % of premium or claims)
5. Profit – can be measured as an ROI, ROE, or % of premium
6. Model office – define the proportion of the block of business that is represented by each model cell

Question 6

Bases used as expected amounts for actual to expected analysis (9)

Answer 6

1. Original pricing assumptions – management likely reviewed these assumptions when the product was being developed, so management expectations may be based on these assumptions
2. Profit targets – this is the bottom line metric that most senior management is interested in
3. Current pricing – may be the most useful measure for inflation sensitive products, since inflation targets are not reliable over the long term
4. Tabular – for DI coverage, a published table is often used for comparison. For DI and LTC, companies with large amounts of data may develop their own internal tables for comparisons

Question 7

ASOP 7 - purposes of cash flow analysis (32)

Answer 7

1. Determination of reserve adequacy
2. Determination of capital adequacy
3. Product development or ratemaking studies
4. Evaluations of investment strategy
5. Financial projections of forecasts
6. Actuarial appraisals
7. Testing of future benefits that may vary at discretion of the insurer (such as dividend scales)

Question 8

ASOP 7 - when to do cash flow testing (32)

Answer 8

1. Situations where it’s needed
   a. There are material asset risks
   b. There are liabilities that have cash flows far out in the future
   c. A company has a new or rapidly growing LOB
   d. Policyholder options are likely to result in antiselection
2. Situations where it isn’t needed
   a. Products with short term liabilities supported by short term assets
   b. Business is not sensitive to changes in economic conditions or interest rates
   c. Risk being evaluated is unanticipated sources of significant claims (like AIDS and asbestos)

Question 9

ASOP 7 - cash flow analysis documentation required (34)

Answer 9

1. Whether any prior analysis were relied on
2. Purpose of analysis and the risks analyzed
3. Type of analysis performed (such as cash flow testing)
4. Results of analysis
5. Actuary’s conclusions or recommendations
6. Conclusions or recommendations related to sensitivity testing
7. Data, assumptions, and methods used

Question 10

ASOP 22 - methods used for asset adequacy analysis (37)

Answer 10

1. Cash flow testing – is appropriate where cash flows of existing assets and liabilities may vary under different economic or interest-rate scenarios
2. Gross premium reserve test – may be appropriate where the policy and other liability cash flows are sensitive to moderately adverse deviations in the actuarial assumptions
3. Demonstration of extreme conservatism – when the degree of conservatism in the liabilities is so great that moderately adverse deviations are covered, then a demonstration of this conservatism is sufficient
4. Demonstration that risk are not subject to material variation – for products that have risks that aren’t subject to material variation, it is sufficient to demonstrate this fact and show that moderately adverse deviations are covered
5. Risk theory techniques – for products with short term liabilities supported by short term assets, it may be more appropriate to measure moderately adverse deviations using risk theory techniques
6. Loss ratio methods – these may be appropriate when the cash flows are of short duration