Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Workshop: Quantitative Finance Day Program at a glance May 5, 2012 (Saturday), 9am - 6pm, Venue: Room No. 2505, Academic Building 2, City University of Hong Kong Welcoming Remarks 9:05am - 9:15am Xunyu ZHOU (CUHK) Session I 9:15am - 9:45am 9:45am - 10:15am 10:15am - 10:45am 10:45am - 11:15am Xiren CAO (SJTU China) Xin GUO (UC Berkeley) Min DAI (NUS Singapore) Break Session II Lunch Session III 11:15am - 11:45am Alex KOLB (Nomura) 11:45am - 12:15pm Duan LI (CUHK) 12:15pm - 12:45pm Lixin WU (HKUST) 12:45pm - 2:00pm 2:00pm - 2:30pm 2:30pm - 3:00pm 3:00pm - 3:20pm Andrew CARVERHILL (HKU) Benny HON (CityU) 3:20pm - 3:50pm 3;50pm - 4:20pm 4:20pm - 4:30pm Zuoquan XU (PolyU) Xiaowei ZHANG (HKUST) Session V 4:30pm 4:50pm 5:10pm 5:30pm 4:50pm 5:10pm 5:30pm 5:50pm Jie SHEN (CityU) Joseph SUNG (HKU) Yingda SONG (HKUST) Tat Wing WONG (CUHK) Closing Remarks 5:50pm - 6:00pm Yuekuen KWOK (HKUST) Break Session IV Break - 1 Titles and Abstracts of Presentations May 5, 2012 (Sat), Room. 2505, Academic Building 2, CityU HK 9:05am - 9:15am, Welcoming Remarks by Xunyu ZHOU, Department of Systems Engineering & Engineering Management, Chinese University of Hong Kong Session I: 9:15am - 10:45am, Chaired by Yuekuen KWOK, Department of Mathematics, University of Science and Technology of Hong Kong 9:15am - 9:45am, Xiren CAO, Department of Finance, Antai school of Business and management, Shanghai Jiaotong University Title: Analysis of Non-linear Behavior - A Sensitivity-Based Approach Abstract: One of the important issues in behavioral analysis is that the law of iterated expectation is lost due to the distortion in performance probability. The standard dynamic programming fails to work in this area. In this talk, we first give a brief review of the different approaches in stochastic optimization and give an overview of the area with a sensitivity-based point of view. Then we apply this sensitivitybased approach, a powerful alternative to dynamic programming, to solve the portfolio management problem in an environment with probability distortion. We show that after changing the underlying probability measure the distorted performance becomes locally linear, and thus the derivative of the distorted performance is simply the expectation of the sample path based derivative of the performance under this new measure, which can be obtained by perturbation analysis. We also provide simulation algorithms for the derivative of distorted performance and hence a gradient-based search algorithm for the optimal policy. We apply this approach to the optimal initial allocation problem with the distorted performance probability and obtained a closed form optimal policy, which is consistent with the results of Jin and Zhou. We expect that this approach is generally applicable to optimization in other non-linear behavioral analysis. 9:45am - 10:15am, Xin GUO, Department of IEOR, UC Berkeley, CA Title: Economic default and Arcsine Law Abstract: This paper develops a structural credit risk model to characterize the difference between the economic and recorded default times 2 for a firm. Recorded default occurs when default is recorded in the legal system. The economic default time is the last time when the firm is able to pay off its debt prior to the legal default time. It has been empirically documented that these two times are distinct (see Guo, Jarrow, and Lin (2008)). In our model, the probability distribution for the time span between economic and recorded defaults is analyzed, and is shown to follow a mixture of Arcsine Laws for some special cases. This is consistent with the results contained in Guo, Jarrow, and Lin. In addition, we show that the classical structural model is a limiting case of our model as the time period between debt repayment dates goes to zero. As a corollary, we show how the firm value process’s parameters can be estimated using the tail index and correlation structure of the firm’s return. 10:15am - 10:45am, versity of Singapore Min DAI, Department of Mathematics, National Uni- Title: Portfolio Selection with Transaction Costs and Multiple Risky Assets Abstract: We consider the optimal consumption and investment with transaction costs on multiple assets, using constant absolute risk aversion utility for the investor and multi-dimensional geometric Brownian motion for the prices of risky assets. We characterize the optimal investment strategy and in particular prove by rigorous mathematical analysis that the trading region has the shape that is very much needed for the trading strategy to be well-defined, e.g. in the two risky asset case, the no-trading region has four corners, the intersections of different types of trading regions are either horizontal or vertical half lines in case of two assets. In contrast, the existing literature is restricted to either single risky asset or multiple uncorrelated risky assets. This work is joint with Xinfu Chen. 10:45am - 11:15am, Break Session II: 11:15am - 12:45am, Chaired by Ken YAN, Nomura Asset Management Hong Kong Limited 11:15am - 11:45am Limited Alex KOLB, Nomura Asset Management Hong Kong Title: Hybrid Derivatives and Correlation: practical and theoretical issues for a robust and consistent framework 3 Abstract: We discuss problems encountered in practice when defining correlations in a derivative pricing framework incorporating multiple assets. Product areas that are susceptible to these considerations are for example derivative payouts that depend on the co-movements of interest rates and equity, or FX linked products that show sensitivities to stochastic interest rates. The recent focus of investment banks on counterparty risk within a portfolio context would be another example where correlation between multiple assets and asset classes are adamant. The lack of reliable market data for various inputs in these kind of derivative products makes a simple but robust choice important. We propose a method how this can be achieved, even within the context of high dimensional (multi factor) models. 11:45am - 12:15pm, Duan LI, Department of Systems Engineering & Engineering Management, Chinese University of Hong Kong Title: Discrete-time behavioral portfolio selection under prospect theory Abstract: We formulate and study a general multi-period behavioral portfolio selection model under Kahneman and Tversky’s prospect theory, featuring an S-shaped value function. Unlike the classical Expected Utility maximization model, a behavioral portfolio model could be easily ill-posed, as He and Zhou (2011) already noticed in their single-period model. Hence, we first discuss the ill-posedness issue and identify the conditions for the well-posedness under a multi-period framework. Under these conditions, we derive then the semi-analytical optimal policy and show that this optimal behavioral portfolio policy still takes a linear feedback form. 12:15pm - 12:45pm, Lixin WU, Department of Mathematics, University of Science and Technology of Hong Kong, Title: Post Crisis Modeling of Interest Rate Derivatives with Credit liquidity and Funding Risks Abstract: We have established a new framework for derivatives pricing under credit, liquidity and funding risks. This framework puts derivative pricing after the financial tsunami back to the track of arbitrage pricing. For interest-rate derivatives, we have extended the affine term structure models and LIBOR market models to accommodate credit and funding risks, in addition to market risks. 4 12:45pm - 2:00pm, lunch Session III: 2:00pm - 3:00pm, Chaired by Qiang ZHANG, Department of Mathematics, City University of Hong Kong 2:00pm - 2:30pm, Andrew CARVERHILL, School of Economics and Finance, Univeristy of Hong Kong Title: Capital Structure Dynamics - Debt, Equity, Liquidity, Dividends and Growth Abstract: We present our recent dynamic model of the firm’s capital structure. The model can reflect a number of empirical regularities, including realistic credit spreads, and liquidity (cash) holdings that do not increase with profitability. We also discuss the issues of optimal leverage, dividend behavior, growth and the “asset substitution problem”, in relation to the model. In the model, liquidity is held not to facilitate investment, but to help the firm survive hard times. We discuss empirical evidence for this and the related measure of the “cash flow (profitability) sensitivity of cash (liquidity)”. 2:30pm - 3:00pm, Benny HON, Department of Mathematics, City University of Hong Kong Title: Kernel-Based Approximation Method for Option Pricing Abstract: In this talk, we present the application of meshless computational method based on the use of kernel-based functions for solving various option pricing problems. Properties of some special kernels such as radial basis functions; harmonic kernels; fundamental and particular solutions; and Greens functions will be discussed. The method requires only a set of nodes in the domain and on the boundary from which all governing equations are solved in strong formulation without the need of tedious meshing and numerical integration. This makes the method advantageous in solve free-boundary type problems such as American options pricing and singularity type problems such as default barrier options pricing. 3:00pm - 3:20pm, Break Session IV: 3:20pm - 4:20pm, Chaired by Hailiang YANG, Department of Statistics & Actuarial Science, University of Hong Kong 5 3:20pm - 3:50pm, Zuoquan XU, Department of Applied Mathematics, Hong Kong Polytechnic University Title: Optimal Stopping under Probability Distortion Abstract: Buy-and-hold, cut-loss-or-take-profit, cut-loss-and-let-profitrun are widely used trading strategies in practice. However, there is no an existing model could explain all of them at once in the literature. It is also well-known that people tend to apply pre-committed strategies when making decisions, however, will change their mind later. In another words, the trading strategies are not time-consistent. By incorporating probability distortion, we formulate an optimal stopping model to derive all the mentioned as well as other trading strategies at once and give an explanation why people use time-inconsistent strategies. 3;50pm - 4:20pm, Xiaowei ZHANG, Department of Mathematics, University of Science and Technology Title: Portfolio Default Losses: Typical and Atypical Behaviors Abstract: Affine point processes are currently widely used in credit risk for modeling portfolio default losses. More specifically, we consider a portfolio exposed to credit risk and partition the constituent firms into several homogeneous groups. The default dynamics of each group is governed by a systematic risk and an idiosyncratic risk factor. All the risk factors form a multivariate affine jump-diffusion process. The framework we consider here incorporates both top-down and bottomup formulations, which are popular approaches in credit risk modeling. In addition, self-excitation and cross-excitation are introduced in our framework in order to capture the clustering feature of defaults. In this talk, we will present the long-term characteristics of portfolio default losses by analyzing the asymptotic behaviors of affine point processes in the large time-horizon asymptotic regime. The first half of the talk will address the typical behaviors of affine point processes, which are characterized by a central limit theorem (CLT). We establish such a CLT by constructing a local martingale and leveraging the CLT for local martingales. Thanks to the affine structure, both the asymptotic average as well as the asymptotic variance can be calculated explicitly. Numerical results show that the Gaussian approximation for the distribution of default losses is fairly accurate, except for the tails. 6 Motivated by the perspective of risk management, which particularly concerns rare but significant events, we will discuss the atypical behaviors of affine point processes in the second half of the talk and characterize them in terms of a large deviation principle, which is proven by applying the Gartner-Ellis theorem. The key step is to calculate the limiting cumulant generating function of the affine point process. Again, thanks to the affine structure, the calculation can be conducted explicitly. Note that the large deviations approximation is only accurate at the logarithmic level. In order to compute the rare event probabilities more precisely, we resort to Monte Carlo simulation but its efficiency heavily depends on the rarity of the event. Using the large deviation result, we develop an importance sampling algorithm for computing the probability of large default losses and prove that it is asymptotically optimal. Numerical results show that our importance sampling algorithm achieves significant variance reduction. 4:20pm - 4:30pm, Break Session V: 4:30pm - 5:50pm, Chaired by Qingshuo SONG, Department of Mathematics, City University of Hong Kong 4:30pm - 4:50pm, Jie SHEN, Department of Mathematics, City University of Hong Kong Title: Saddle Points of Discrete Markov Zero-Sum Game with Stopping Abstract: We study the sufficient conditions for the existence of a saddle point of time-dependent discrete Markov zero-sum game up to a given stopping time. The stopping time is allowed to take either a finite or an infinite non-negative random variable with its associated objective function being well-defined. The result enables us to show the existence of the saddle points of discrete games constructed by Markov chain approximation of a class of stochastic differential games. This is a joint work with Xun Li and Qingshuo Song. 4:50pm - 5:10pm, Joseph SUNG, Department of Statistics & Actuarial Science, University of Hong Kong Title: A Nonzero-Sum Stochastic Differential Reinsurance Game With Mixed Regime Switching 7 Abstract: In this work, we study a class of nonzero-sum stochastic differential reinsurance games between two insurance companies. Each insurance company is assumed to maximize the difference of the opponent’s terminal surplus from that of its own by properly arranging its reinsurance schedule. The surplus process of each insurance company is modeled by a mixed regime-switching Cramer-Lundberg approximation, that is, a diffusion risk process with coefficients being modulated by both a continuous-time finite-state Markov Chain and another diffusion process. By adopting the stochastic HJB equation approach, we provide a verification theorem to characterize the underlying Nash Equilibrium. In particular, under the most commonly used exponential utility situation, we further obtain the explicit optimal reinsurance schedule. The economic interpretation of the optimal strategy closely matches with the common practice of insurance companies in realty. (This is a joint work with A. Bensoussan, S.C.P. Yam and S.P. Yung.) 5:10pm - 5:30pm, Yingda SONG, Department of IELM, University of Science and Technology Title: Exact Simulation of the SABR Model Abstract: The stochastic alpha-beta-rho (SABR) model becomes popular in the financial industry because it is capable of providing good fits to various types of implied volatility curves observed in the marketplace. However, no analytical solution to the SABR model exists that can be simulated directly. Although we can apply discretization methods to simulate security price under the SABR model, they introduce discretization bias to the simulation results. Therefore, we may have to use a large number of time steps to reduce the bias to an acceptable level. This paper proposes a Monte Carlo method for the exact simulation of the forward price and its volatility under the SABR model. Primary difficulties involved in our exact simulation scheme are how to simulate two random variables whose distributions can be expressed in terms of the Hartman-Watson and the noncentral chi-squared distribution functions, respectively. Two novel simulation schemes are proposed to achieve numerical accuracy, efficiency, and stability. One stems from numerical Laplace inversion and Asian option literature, and the other is based on recent developments in evaluating the noncentral chi-squared distribution functions in a robust way. Numerical examples demonstrate that our method is fast and accurate under various market environments. 8 5:30pm - 5:50pm, Tat Wing WONG, Department of Statistics, Chinese University of Hong Kong Title: Managing Mortality Risk with Longevity Bonds When Mortality Rates Are Cointegrated Abstract As a hedging vehicle for insurance companies to manage their mortality risks, longevity bonds are linked to a selected mortality index. We investigate the dynamic mean-variance hedging problem of an insurer using longevity bonds. Insurance liabilities are modelled by a doubly stochastic compound Poisson process with the mortality rate being correlated and cointegrated with the index mortality rate. We solve this dynamic hedging problem using theory of backward stochastic differential equations. Our theory shows that cointegration materially affects the optimal hedging strategy on top of correlation. The effect of cointegration is independent of the risk preference of insurers. Explicit solutions of the optimal hedging strategy are derived for both cointegrated stochastic mortality models with constant volatilities and with state-dependent volatilities. (This is a joint work with M.C. Chiu and H.Y. Wong) 5:50pm - 6:00pm, Closing Remarks by Yuekuen Kwok, Department of Mathematics, University of Science and Technology 9