Longevity Risk Management: Models and Hedging Strategies

Abstract

Longevity risk management is becoming increasingly important in the pension and life insurance industries. The unexpected mortality improvements observed in recent decades are posing serious concerns to the financial stability of defined-benefit pension plans and annuity portfolios. It has recently been argued that the overwhelming longevity risk exposures borne by the pension and life insurance industries may be transferred to capital markets through standardized longevity derivatives that are linked to broad-based mortality indexes. To achieve the transfer of risk, two technical issues need to be addressed first: (1) how to model the dynamics of mortality indexes, and (2) how to optimize a longevity hedge using standardized longevity derivatives. The objective of this thesis is to develop sensible solutions to these two questions.

In the first part of this thesis, we focus on incorporating stochastic volatility in mortality modeling, introducing the notion of longevity Greeks, and analysing the properties of longevity Greeks and their applications in index-based longevity hedging. In more detail, we derive three important longevity Greeks---delta, gamma and vega---on the basis of an extended version of the Lee-Carter model that incorporates stochastic volatility. We also study the properties of each longevity Greek, and estimate the levels of effectiveness that different longevity Greek hedges can possibly achieve. The results reveal several interesting facts. For example, we found and explained that, other things being equal, the magnitude of the longevity gamma of a q-forward increases with its reference age. As with what have been developed for equity options, these properties allow us to know more about standardized longevity derivatives as a risk mitigation tool. We also found that, in a delta-vega hedge formed by q-forwards, the choice of reference ages does not materially affect hedge effectiveness, but the choice of times-to-maturity does. These facts may aid insurers to better formulate their hedge portfolios, and issuers of mortality-linked securities to determine what security structures are more likely to attract liquidity.

We then move onto delta hedging the trend and cohort components of longevity risk under the M7-M5 model. In a recent project commissioned by the Institute and Faculty of Actuaries and the Life and Longevity Markets Association, a two-population mortality model called the M7-M5 model is developed and recommended as an industry standard for the assessment of population basis risk. We develop a longevity delta hedging strategy for use with the M7-M5 model, taking into account of not only period effect uncertainty but also cohort effect uncertainty and population basis risk. To enhance practicality, the hedging strategy is formulated in both static and dynamic settings, and its effectiveness can be evaluated in terms of either variance or 1-year ahead Value-at-Risk (the latter is highly relevant to solvency capital requirements). Three real data illustrations are constructed to demonstrate (1) the impact of population basis risk and cohort effect uncertainty on hedge effectiveness, (3) the benefit of dynamically adjusting a delta longevity hedge, and (3) the relationship between risk premium and hedge effectiveness.

The last part of this thesis sets out to obtain a deeper understanding of mortality volatility and its implications on index-based longevity hedging. The volatility of mortality is crucially important to many aspects of index-based longevity hedging, including instrument pricing, hedge calibration, and hedge performance evaluation. We first study the potential asymmetry in mortality volatility by considering a wide range of GARCH-type models that permit the volatility of mortality improvement to respond differently to positive and negative mortality shocks. We then investigate how the asymmetry of mortality volatility may impact index-based longevity hedging solutions by developing an extended longevity Greeks framework, which encompasses longevity Greeks for a wider range of GARCH-type models, an improved version of longevity vega, and a new longevity Greek known as `dynamic delta'. Our theoretical work is complemented by two real-data illustrations, the results of which suggest that the effectiveness of an index-based longevity hedge could be significantly impaired if the asymmetry in mortality volatility is not taken into account when the hedge is calibrated.