节点文献
具有交易成本的最优投资组合及极限定理
The Optimal Portfolio and Limit Theorem with Transaction Costs
【作者】 徐云;
【作者基本信息】 新疆大学 , 应用数学, 2004, 博士
【摘要】 马克思说过:一种科学只有在成功地运用数学时,才算达到了真正完善的地步. 金融学的定量研究越来越引起人们的重视,尤其是20世纪90年代亚洲经济危机以后. 数学金融学作为现代金融学的核心自20世纪90年代以来发展迅猛. 用数学方法研究金融问题可追述到上世纪初. 1900年, 法国学者路易斯 ·巴谢利耶(Louis Bachelier, 1900)发表了名为《投机理论》的博士论文[5],首次对 Brown 运动给出了精确的数学描述. 1952年,Markowitz 发表了他的博士论文[56], 提出“均值方差理论”,标志着现代金融学的开始,也是公认的现代金融学经历的第一次革命. 随后,Sharpe(1964) [78] , Lintner(1965)[53] 和 Merton (1972) [61] 进一步拓展了 Markowitz 的工作,提出“资本资产定价模型”(CMPA). 20世纪60年代,Samuelson(1965) [75] 和Fama(1965)[32] 的“市场有效性假设”是对市场完备性的描述. 1973年,Black, Scholes[10] 和 Merton [62] 发表“期权定价公式”是现代金融学的二次革命的标志.两次革命推动了现代金融学发展. 投资者面临的主要问题是如何在非确定性的环境中使其财富最大化.Markowitz 的“均值方差”理论将人们期望找到“最好”股票的想法引导到对风险和收益的量化和均衡上. 不久,均值方差模型扩拓展到用投资组合的期望回报作为投资回报,并用组合的期望回报的方差作为投资风险. 由于证券组合期望回报的变化,几乎所用投资组合优化都是对现有组合的修改. 具有交易费用的投资组合管理几乎多数都集中于 Merton 问题. 本文考虑将风险函数一般化后具有交易费用的单时段最优投资组合存在的必要条件, 多时段最优投资组合序列的极限问题以及一般随机 LQ 框架下完全市场中连续时间最优投资组合和消费问题. 第一章,我们利用时间序列 markov 链组合预测模型分析股票价格指数与国家政策的关系,并用上证数据进行实证分析,得出政策因素对股价影响程度的结论. vi<WP=9>新疆大学博士学位论文 徐云第二章,建立了单时段具有交易费用的最优投资组合问题的凸规划模型,接着讨论了模型最优解的存在条件和最优解的等价关系. 特别地,将风险函数特殊化为投资组合期望收益率的方差,模型即变为具有交易费用的均值方差模型,此时具体得到了模型的最优解的条件.第三章,考虑了与上章相同的多时段问题,得到了多时段具有交易费用的最优投组合的凸规划模型,接着研究了最优投资组合的序列的极限与相应的投资组合收益序列极限的关系.第四章,我们考虑了具有信息作用时具有交易费用的多时段投资组合的凸规划问题,并解释了在实际应用中的经济含义.最后一章,建立了一般随机 LQ 框架下连续时间最优投资组合选择和消费的模型,并通过将原问题转化为标准随机 LQ 问题,利用随机 Riccati 方程得到问题的解析形式的最优解. 说明在风险管理中,给定了收益水平,能够找到最优的交易策略使风险最小化.由于考虑了交易成本和消费,所得结果在投资者的风险投资管理中具有实际意义.
【Abstract】 The major problem facing investors has always been the maximizationof wealth in a world of uncertainty. Portfolio selection is how to allocatewealth among alternative assets of a basket. The mean-variance approach byMarkowitz provides a fundamental basis for portfolio construction in a singleperiod. The most important contribution of this model is that it quanti?es therisk by using the variance, which enables invertors to seek the highest returnafter specifying their acceptable risk level. E?orts in extending, the core of the Markowitz mean-variance model is totake the expected return of a portfolio as the investment return and the vari-ance of expected returns of a portfolio as the investment risk. Due to changein the expectation of the future returns of securities, most of the applicationsof portfolio optimization involve the revision of an existing portfolio. Therehave been numerous e?orts to illuminate the role of transaction costs portfoliomanagement, much of the work having been focused on Merton’s problem. This thesis focuses on there existence condition of the optimal solution ofwhich convex programming model of portfolio select and it’s limit of portfolioselection problem with transaction costs based on generalization risk function,and discussed the continuous-time portfolio selection with consumption understochastic linear-quadratic framework. The organization is as follows. The ?rst chapter, we construct a combination predictive model of timeseries markov chain, which has been tested on real data from the ShanghaiStock Market. Using several ?tting predict methods in the empirical analysis. The second chapter, a convex programming model with respect portfolioselect with transaction costs in single period was present, then we provedthe optimal solution’s existence condition and the equivalence of this model.Finally, we illustrate a special example for the optimal solution. viii<WP=11>新疆大学博士学位论文 徐云 The third chapter, we discuss the relationship between the limitation of theoptimal solution sequence and corresponding expectation of relate return se-quence of portfolio in multiperiod, prove the limit Theorem of above sequences. In fourth chapter, the limit Theorem of optimal portfolio in multiperiodwith information was present. Then given the economic interpret of somecases. The last chapter concerned with a continuous-time portfolio selection modelon optimal control. The objective is to minimize the variance of the termi-nal wealth for a given expectation return level. We obtain the optimal so-lution of the original problem located via the solution of a stochastic linear-quadratic(LQ).
【Key words】 Transaction costs; Portfolio; model; Optimal solution; Limit theorem; Optimal control.;