Description: The speaker will discuss the training of neural networks to compute future Greeks of liabilities (or options) without the use of nested stochastics.

Description: This article departs from the usual GMWB pricing approach by assuming that the policyholder is always smart enough to act in his/her own best interest.

Description: This article provides some insights in how to bring risk management into product development for VA guarantees.

Description: The proposed revisions to the variable annuity statutory framework and the FASB targeted improvement for long duration contract, motivate life insurance companies to re-evaluate their hedging strategy. Recently proposed revisions of AG 43/VM-21 and C3 Phase II, reward companies that have a clearly defined hedging strategy, if the companies can demonstrate that the valuation model and capital model can estimate real hedge performance accurately. The speakers of this session will discuss the hedging implication of the proposed AG 43/VM-21 and C3 Phase II framework. Additionally, methodology computational requirements, back-testing results under different economic scenarios and the resulting hedge credits of different hedging strategies will be compared, including explicit dynamic hedge and implicit dynamic hedge. The speakers will also touch on the impact of rebalance frequency on the performance of dynamic hedge. Case studies will be presented to provide guidance on practical implementation.

Description: This article focuses on how a risk appetite framework can help a company manage the “three-headed monster” and why a risk appetite framework is critical to effective business and risk management at both the corporate and product levels.

Description: Regulatory change and increased focus on internal risk management are driving a renewed interest in model efficiency. In this session, we will review several new and existing tools, that show significant promise for advanced modeling of equity linked products. These include: • Clustering • Use of unique--and potentially a different number of--scenarios for each policy • Semi Monte Carlo (SMC) – this is a proprietary technique that is similar to a combination of Monte Carlo and numeric integration • Using a neural network approach to optimize some of the more complex and manual aspects of clustering and improve the fit of the clustered model • Combining SMC with the use of unique scenarios per policy

Description: Segregated Funds Seminar Illuminates Equity Guarantees Risks by G. Thomas Mitchell from The Financial Reporter, February 2000, Issue No. 41.