Csp Q36

Front 1

QBS 36: 1-3 of 21
Uses of Forecasts
(1-3 of 3)

Back 1

1) Determine plan's future finances
2) Reduce uncertainty of possible future events
3) Long-term strategic planning
4) Budgeting
5) Avoid unpleasant surprises
6) Gain a competitive edge
7) Better understanding relationship between assets and liabilities
8) Determine impact of events such as:
a) Early retirement window
b) Chance in demographics and its impact on plan's financial position
c) Changes in economy
d) Changing an asset mix
e) Plan changes
f) Legislative changes
g) Funding method change
h) Funding policy changes
9) Applying/developing immunization and convexity theories
10) Determine solvency liability
11) Estimate benefit disbursement and liquidity requirements
12) Determine supportability of multiemployer plans and social security programs
13) Gain better insight regarding
a) Cash flows
b) Asset accumulations
c) Salaries
d) Population size

Front 2

QBS 36: 4 of 21
Deterministic Forecast

Back 2

- Assumptions don't vary once selected

Front 3

QBS 36: 5 of 21
Limitations of Deterministic Projections

Back 3

1) Assumptions will never be fully realized
2) Difficult to do any true sensitivity analysis
3) Plan sponsor view results as cast in stone

Front 4

QBS 36: 6 of 21
Purpose of a stochastic model

Back 4

- Make sponsor aware of the potential range of outcomes over the forecast period, while indicating the probabilities and volatility involved

Front 5

QBS 36: 7 of 21
Stochastic Capital Market Assumptions (CMA) used for assets, salary scale, discount rate, include

Back 5

1) Expected future inflation
2) Expected future real return based on portfolio
3) Variance or SD of the expected returns and inflation
4) Correlation between real returns and inflation and between real returns on different asset classes

Front 6

QBS 36: 8 of 21
Stochastic projections include the following steps

Back 6

1) Stochastic asset return modeling
2) Stochastic liability modeling
3) Calculation of results for each year, for each trial
4) Ranking of results and production of confidence intervals
5) Preparation of results

Front 7

QBS 36: 9 of 21
Advantages of stochastic over deterministic

Back 7

1) Results provided in confidence intervals instead of single answers (average risk)
2) Range of results convey one single answer should not be relied upon with certainty
3) True sensitivity can be performed
4) Can reflect difference in risk tolerance between plan sponsors
5) Facilitates communication with investment managers
6) Shows relationship between investment performance, contributions, expense and benefit security

Front 8

QBS 36: 10 and 11 of 21
Steps of projection study
(1 and 2 of 2)

Back 8

1) Discuss the scope of the project
2) Determine assumptions for projecting the population forward
3) Collect data
4) Produce liability stream for deterministic part of the projection
5) Combine results with projected asset returns to produce valuation results
6) Present results to client
- Certain items may be identified for further stochastic study
7) Determine assumptions and scenarios for stochastic projection
8) Perform stochastic projections
9) Present the stochastic results to the client

Front 9

QBS 36: 12 of 21
Valuation versus Projection Assumptions

Back 9

1) Valuation based on population as a fixed date
2) Valuation based on long-term assumptions
3) Projection assumptions referred to as real world assumptions (reflect short-term expectations)

Front 10

QBS 36: 13 and 14 of 21
Potential Differences between Valuation and Projection Demographic Assumptions
(1 and 2 of 2)

Back 10

1) Mortality and Disability
- Typically val and projection use same assumptions
2) Withdrawal
- May be the same if short-term expectations are the same as long-term
- Projection should reflect economic conditions like poor economy and layoffs
- If use ultimate rates for val then should consider select and ultimate for projection
3) Retirement
- Val may use a single retirement age
- Projection should use expected rates since benefit payments affect assets

Front 11

QBS 36: 15 and 16 of 21
Potential Differences between Valuation and Projection Economic Assumptions
(1 and 2 of 2)

Back 11

1) Salary Increase Assumptions
- Projection assumptions get population from one valuation date to the next
- Once they are there the val assumptions determine the liabilities and NC
- Usually only varies by age for val
- For projection salary will often vary by calender year
2) Social Security and Benefit Limitation Assumptions
- Should be consistent with inflation and productivity assumptions from projection and val
3) Investment Return
- Fixed for val
- May change year to year for projection

Front 12

QBS 36: 17 of 21
Other Assumption Considerations

Back 12

- Optional form utilization should be considered for projection since affects benefit payment stream
- Should also be considered for val if not actuarially equivalent
- May reflect COLA for projection but may not be allowed to for valuation

Front 13

QBS 36: 18 of 21
Investment assumptions needed for projection

Back 13

1) Total return
- May be constant, but typically vary for projection
- Only need total return if AVA or MRVA = MV
2) Portfolio return attributable to interest realized and unrealized gains/losses differently
3) Portfolio turnover
- Needed if activity material and projected AVA splits the gains/losses into realized and unrealized appreciation

Front 14

QBS 36: 19 and 20 of 21
Deterministic projections, should check the following
(1 and 2 of 2)

Back 14

1) Liability streams
- Liability and NC increases should be smooth
- NC as a % of pay should be stable
- Liability and benefit payments should increase smoothly if new retirees steady
- Large increase in inactive should be offset by decrease in active liability
- Should not have gain/loss if projection assumptions same as valuation assumptions
2) Check base case results
- Verify initial year results match valuation report
3) Typically several scenarios run in projection
- May change one and compare for reasonableness

Front 15

QBS 36: 21 of 21
Stochastic projections,should check the following

Back 15

1) Check a detailed trace of the results for the first few trials
2) Check confidence intervals
- Check tails closely