No description available.

Thesis (M. Phil.)--University of Hong Kong, 2010. / Includes bibliographical references (leaves 73-77). Also available in print.

published_or_final_version / Statistics and Actuarial Science / Master / Master of Philosophy

本研究主要探討附保證最低提 (Guaranteed Minimum Withdrawal Benefits, GMWB)的變額(Variable Annuity, VA) 在隨機模型下之定價。保證最低提是變額的一種附加約 (rider) 並在市場下跌的情況下為變額持有人提供保障。它保證持有人在合約期內的總提少於一個預先訂的額，而變額的投資表現。一般，這個保證額相等於變額的初始投資額。本研究的融模型假設投資標的基價格符合對維過程 (exponential Lévy process)，而隨機則符合由維過程驅動的瓦西克模型 (Vasiček model)。融模型中的個維過程的相依結構 (dependence structure) 會由維關結構 (Lévy Copula) 描述。這個方法的好處是可描述同型的相依結構。用一個配合維關結構而有效的蒙地卡模擬方法，我們研究在同相依結構及模型下保證最低提的價值變化。在固定的特別情況下，保證最低提的價值能夠透過卷積方法 (convolution method) 而得到半解析解 (semi-analytical solution) 。最後，我們將本研究中的學模型擴展以研究近期出現由保證最低提演化而成的一種保證產品。這個產品名稱為保證終身提 (Guaranteed Lifelong Withdrawal Benefit, GLWB)，而此產品的到期日則與持有人的壽命相關。 / In this thesis, we study the problem of pricing the variable annuity(VA) with the Guaranteed Minimum Withdrawal Benefits (GMWB) under the stochastic interest rate framework. The GMWB is a rider that can be elected to supplement a VA. It provides downside protection to policyholders by guaranteeing the total withdrawals throughout the life of the contract to be not less than a pre-specied amount, usually the initial lump sum investment, regardless of the investment performance of the VA. In our nancial model, we employ an exponential L´evy model for the underlying fund process and a Vasiček type model driven by a L´evy process for the interest rate dynamic. The dependence structure between the two driving L´evy processes is modeledby the L´evy copula approach whichis exible to model a wide range of dependence structure. An effcient simulation algorithm on L´evy copula is then used to study the behavior of the value of the GMWB when the dependence structure of the two L´evy processes and model parameters Vry. When the interest rate is deterministic, the value of the GMWB can be solved semi-analytically by the convolution method. Finally, we extend our model to study a recent variation of GMWB called Guaranteed Life long Withdrawal Benefits (GLWB) in which the maturity of the GLWB depends on the life of the policyhodler. / Detailed summary in vernacular field only. / Chan, Wang Ngai. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2012. / Includes bibliographical references (leaves 115-121). / Abstracts also in Chinese. / Abstract --- p.i / Acknowledgement --- p.iv / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Variable Annuity & Guaranteed Minimum Withdrawal Benefit --- p.1 / Chapter 1.2 --- Literature Review --- p.4 / Chapter 1.3 --- Financial Model for GMWB --- p.7 / Chapter 2 --- L´evy Copulas and the Simulation Algorithm --- p.12 / Chapter 2.1 --- Definitions and Theorem --- p.15 / Chapter 2.2 --- Examples of L´evy Copulas --- p.19 / Chapter 2.2.1 --- Independence case --- p.19 / Chapter 2.2.2 --- Complete Dependence --- p.20 / Chapter 2.2.3 --- The Clayton L´evy Copula --- p.21 / Chapter 2.3 --- Simulation algorithm for two-dimensional dependent L´evy process --- p.22 / Chapter 3 --- Model Formulation for GMWB --- p.26 / Chapter 3.1 --- Financial Model for GMWB --- p.27 / Chapter 3.2 --- Underlying Fund of VA and the Interest Rate --- p.30 / Chapter 3.3 --- A Special Case of Deterministic Interest Rate --- p.34 / Chapter 4 --- Numerical Implementation --- p.38 / Chapter 4.1 --- The Clayton L´evy Copula --- p.39 / Chapter 4.2 --- The Underlying Fund and the Interest Rate Processes --- p.42 / Chapter 4.3 --- Kendall’s Tau Coefficient --- p.47 / Chapter 4.4 --- The GMWB Option Value --- p.49 / Chapter 4.4.1 --- Control Variate for Simulation --- p.49 / Chapter 4.4.2 --- Simulation Results --- p.51 / Chapter 4.5 --- Deterministic Interest Rate --- p.52 / Chapter 5 --- GMWB Pricing Behavior --- p.56 / Chapter 5.1 --- L´evy model for the underlying fund --- p.57 / Chapter 5.1.1 --- The Skewness --- p.57 / Chapter 5.1.2 --- The Kurtosis --- p.65 / Chapter 5.2 --- The Vasiček model driven by L´evy process --- p.73 / Chapter 5.2.1 --- The Volatility Parameter ôV --- p.73 / Chapter 5.2.2 --- The Mean Reverting Parameter aV --- p.77 / Chapter 5.3 --- Dependence between the underlying fund and rate processes --- p.81 / Chapter 5.3.1 --- The jump direction dependence parameter n{U+1D9C} --- p.83 / Chapter 5.3.2 --- The jump magnitude dependence parameter θ{U+1D9C} --- p.90 / Chapter 6 --- GMWB for Life --- p.96 / Chapter 6.1 --- Model Formulation --- p.98 / Chapter 6.1.1 --- Mortality model --- p.99 / Chapter 6.1.2 --- Financial Model for GLWB --- p.101 / Chapter 6.2 --- GLWB product from John Hancock --- p.103 / Chapter 6.3 --- GLWB Pricing Behavior --- p.104 / Chapter 6.3.1 --- The correlation effect --- p.106 / Chapter 7 --- Conclusion --- p.108 / A Proofs --- p.113 / Chapter A.1 --- Proof of Equation 3.1 --- p.113 / Chapter A.2 --- Proof of Equation 3.3 --- p.114 / Bibliography --- p.115

Several of the more complex optimization problems in finance can be characterized as impulse control problems. Impulse control problems can be written as quasi-variational inequalities, which are then solved to determine the optimal control strategy. Since most quasi-variational inequalities do not have analytical solutions, numerical methods are generally used in the solution process. In this thesis, the impulse control problem framework is applied to value two complex long-term option-type contracts. Both pricing problems considered are cast as impulse control problems and solved using an implicit approach based on either the penalty method or the operator splitting scheme. The first contract chosen is an exotic employee stock option referred to as an infinite reload option. This contract provides the owner with an infinite number of reload opportunities. Each time a reload occurs, the owner pays the strike price using pre-owned company shares and, in return, receives one share for each option exercised and a portion of a new reload option. Numerical methods based on the classic Black-Scholes equation are developed while taking into account contract features such as vesting periods. In addition, the value of an infinite reload option to it's owner is obtained by using a utility maximization approach. The second long-term contract considered is a variable annuity with a guaranteed minimum death benefit (GMDB) clause. Numerical methods are developed to determine the cost of the GMDB clause while including features such as partial withdrawals. The pricing model is then used to determine the fair insurance charge which minimizes the cost of the contract to the issuer. Due to the long maturity of variable annuities, non-constant market parameters expressed through the use of regime-switching are included in the GMDB pricing model.

Several of the more complex optimization problems in finance can be characterized as impulse control problems. Impulse control problems can be written as quasi-variational inequalities, which are then solved to determine the optimal control strategy. Since most quasi-variational inequalities do not have analytical solutions, numerical methods are generally used in the solution process. In this thesis, the impulse control problem framework is applied to value two complex long-term option-type contracts. Both pricing problems considered are cast as impulse control problems and solved using an implicit approach based on either the penalty method or the operator splitting scheme. The first contract chosen is an exotic employee stock option referred to as an infinite reload option. This contract provides the owner with an infinite number of reload opportunities. Each time a reload occurs, the owner pays the strike price using pre-owned company shares and, in return, receives one share for each option exercised and a portion of a new reload option. Numerical methods based on the classic Black-Scholes equation are developed while taking into account contract features such as vesting periods. In addition, the value of an infinite reload option to it's owner is obtained by using a utility maximization approach. The second long-term contract considered is a variable annuity with a guaranteed minimum death benefit (GMDB) clause. Numerical methods are developed to determine the cost of the GMDB clause while including features such as partial withdrawals. The pricing model is then used to determine the fair insurance charge which minimizes the cost of the contract to the issuer. Due to the long maturity of variable annuities, non-constant market parameters expressed through the use of regime-switching are included in the GMDB pricing model.

Policyholder exercise behavior presents an important risk factor for life insurance companies. Yet, most approaches presented in the academic literature – building on value maximizing strategies akin to the valuation of American options – do not square well with observed prices and exercise patterns. Following a recent strand of literature, in order to gain insights on what drives policyholder behavior, I first develop a life-cycle model for variable annuities (VA) with withdrawal guarantees. However, I explicitly allow for outside savings and investments, which considerably affects the results. Specifically, I find that withdrawal patterns after all are primarily motivated by value maximization – but with the important asterisk that the value maximization should be taken out from the policyholders’ perspective accounting for individual tax benefits. To this effect, I develop a risk-neutral valuation methodology that takes these different tax structures into consideration, and apply it to our example contract as well as a representative empirical VA. The results are in line with corresponding outcomes from the life cycle model, and I find that the withdrawal guarantee fee from the empirical product roughly accords with its marginal price to the insurer. I further consider the implications of policyholder behavior on product design. In particular – due to differential tax treatments and contrary to option pricing theory – the marginal value of such guarantees can become negative, even when the holder is a value maximizer. For instance, as I illustrate with both a simple two-period model and an empirical VA, a common death benefit guarantee may indeed yield a negative marginal value to the insurer.

Product risk is important to firms’ enterprise risk management. This dissertation focuses on product risk in the U.S. life insurance and health insurance industries. In particular, we add new dimensions to the measurement of product risk for these industries, and we explore how these industries manage product risk in a context of other enterprise risks. In this dissertation, we identify new product risks, propose new measures, and study the management of these risks. In the life insurance industry, we identify a new type of product risk, the guarantee risk, caused by variable annuities with guaranteed living benefits (VAGLB). We propose a value-at-risk type measure inspired by the risk-based capital C3 Phase II to quantify the guarantee risk. In the health insurance industry, where the degree of uncertainty varies for different types of health insurance policies, we develop four exposure-based risk measures to capture health insurers’ product risks. Then we study how life and health insurers manage product risks (and asset risks) by using capital in the context of other risks and appropriate controls. We add to the literature in the life insurance industry by examining the relationship between capital and risks when the guarantee risk is accounted for. In the health insurance industry, to our knowledge, no similar research on the relationship between capital and risks has been conducted. In view of the current topicality of health insurance, our research therefore adds a timely contribution to the understanding of health insurer risk management in an era of health care reform. Capital structure theories, transaction cost economics, and insurers’ risk-taking behaviors provide the theoretical foundation for our research. As to methodology, we implement standard capital structure models for the life and health insurance industries using data from the National Association of Insurance Commissioners (NAIC) annual filings of life/health insurers and health insurers. Simultaneous equations modeling is used to model life and health insurers’ enterprise risk management. And the estimation is conducted by the generalized estimation equations (GEE). We find that both U.S. life/health insurers and health insurers prudently build up capital as they experience more product risk and asset risk controlling for the other enterprise risks. We also find that life/health insurers may be using derivatives as a partial substitute for capital when managing new product risk caused by VAGLB, the guarantee risk. / text

This dissertation provides a comprehensive and practical analysis of basis risk in the U.S. variable annuity market and examines effective fund mapping strategies to mitigate the level of basis risk while controlling for the associated transaction costs. Variable annuities are personal savings and investment products with long-term guarantees that expose life insurers to extensive financial risks. Liabilities associated with VA guarantees are the largest liability component faced by U.S. life insurers and have raised concerns to VA providers and regulators. And the hedging performance of these guarantee liabilities is impeded by the existence of basis risk. I look into 1,892 registered VA-underlying mutual funds and two VA separate accounts to estimate the basis risk faced by U.S. VA providers at the individual fund level and the separate account level. To evaluate the degree to which basis risk can be mitigated, I consider various proxy instrument sets and assess different variable selection models. The LASSO regression is shown to be most effective at identifying the most suitable (combination of) mapping instruments that minimize basis risk, compared to other test-based and screening-based models. I supplement it with the Sure Independence Screening (SIS) procedure to further limit the number of instruments requested in the hedging strategies, and modify it by introducing the diff LASSO regression to restrict the changes in instrument allocations across rebalancing periods and, therefore, control for transaction costs. I show that VA providers can reduce their exposure to basis risk by applying data analytic techniques in their mapping process, by hedging with ETFs instead of futures contracts, and through diversification at the separate account level. Combining the traditional fund mapping method with the machine learning algorithm, the proposed portfolio mapping strategy is efficient at reducing basis risk in VA separate accounts while controlling for the tractability and transaction costs of the mapping and hedging procedure, and is practical to incorporate newly-developed VA funds, as well as the varying compositions of separate accounts. Overall, this study presents that U.S. VA providers have the ability to mitigate basis risk to a greater extent than the limited literature on this topic has suggested. / Business Administration/Risk Management and Insurance

La présente thèse s'intéresse à trois acteurs du financement des régimes de retraite : le législateur, l'assureur et l’individu. Dans un environnement en proie à un comportement déviant du marché financier et à des évolutions démographiques défavorables, le rôle de ces parties prenantes doit impérativement faire l’objet d’une réévaluation pour relever le défi de la pérennité du financement des retraites. L’étude de la règlementation et de la conception des régimes a été réalisée en intégrant des caractéristiques types du futur paysage des retraites, telles que le poids de plus en plus important du risque assumé par l’individu ou l’éventuelle participation d'investisseurs boursiers dans l’offre de contrats. Les conclusions de cette étude permettent de dégager des orientations en vue de la gestion du risque de longévité pour les individus, une évaluation de l’attrait de l’exposition au risque de longévité pour les investisseurs, des informations sur l’élaboration des contrats pour les assureurs ainsi que des propositions, pour les décideurs politiques, de mesures règlementaires favorisant la durabilité du paysage des retraites. / This dissertation examines three stakeholders in pension finance: the individual, the policymaker, and the pension provider (e.g., an insurer or a pension fund). In a setting beset by unforseen financial market circumstances and demographic changes that disfavor financial security in retirement, a re-evaluation of these stakeholders’ role is necessary. We explore the regulation and design of retirement plans by incorporating features that characterize the future retirement landscape, such as the increasing burden of risk borne by the individual, and the potential involvement of market investors in the provision of retirement contracts. The implications of our findings encompass guidance for individuals in managing longevity risk, evaluation of the appeal of longevity risk exposure to investors, insights on contract design for the insurer, and proposals to the policymaker on regulatory measures that foster a sustainable retirement environment.