Page 3
https://study.cfainstitute.org/app/cfa-institute-program-level-iii-for-august-2024#read/study_task/2562293/factor-modeling-in-liability-relative-approaches-1
FACTOR MODELING IN LIABILITY-RELATIVE APPROACHES
Learning Outcomes
describe the use of investment factors in constructing and analyzing an asset allocation
discuss approaches to liability-relative asset allocation
A factor-based approach for liability-relative asset allocation has gained interest and credibility for several reasons. First, in many applications, the liability cash flows are dependent on multiple uncertainties. The two primary macro factors are future economic conditions and inflation. Many pension payments to beneficiaries will be based on inflation and salary changes over the employees’ work span. A fully hedged portfolio cannot be constructed when the liabilities are impacted by these uncertain factors. Recall that a hedged portfolio can be constructed for a frozen plan with fixed liabilities. For ongoing pension schemes, the best that can be done is to add asset categories to the portfolio that are positively correlated with the underlying driving risk factors, such as inflation-linked bonds. A factor-based approach can be implemented with any of the three liability-relative asset allocation methods discussed above.
EXAMPLE 8
Robustness and Risk Assessment in Liability-Relative Asset Allocation
What types of sensitivity analysis can be evaluated with a multi-period ALM simulation system?
Solution:
To provide estimates of the probability of meeting future obligations and the distribution of outcomes, several types of sensitivity analysis are likely to be performed.
For example, the expected returns could be increased or decreased to evaluate the impact on future contributions to the plan.
Likewise, by analyzing historical events, the investor can estimate the size of losses during crash periods and make decisions about the best asset allocation to protect against these worst-case events. Multiple risk measures over time (temporal risk measures) can be readily included in a simulation system.
Last updated