【作者】 张欣；

【导师】 匡海波；

【作者基本信息】 大连海事大学 ， 工程硕士（专业学位）， 2021， 硕士

【摘要】 国际航运业是一个资本密集且投资风险大的服务性行业,近年来,油轮市场运价波动不断,极大影响了油轮运输企业的经营收益,油轮投资成为了油轮运输企业经营活动最重要的内容,具有很强的战略意义。但是,油轮市场投资金额大、回收期长和风险性高的特点,决定了油轮投资具有很大的复杂性和不确定性,所以合理的资产投资配置不仅会提升公司的投资收益,还能增强企业的抗风险能力,扩大公司船队的规模,提高企业竞争力。因此,油轮投资组合配置的优化异常关键。本文采用与现实更加吻合的稳定分布刻画油轮收益率的偏态、过度峰态以及厚尾的特性,通过尺度和风险价值Va R度量风险;通过偏斜指数提高获得超额收益的概率;通过熵值降低历史数据的影响,避免产生投资资金过于集中的情况;通过偏好度约束有效满足不同投资者的投资需求。将收益率、风险价值、尺度参数、偏斜指数与熵联立,建立了基于稳定分布的“均值-Va R-尺度参数-偏斜指数-熵”多目标油轮投资组合优化模型,同时考虑目标优化的偏好度设置,完善了传统投资组合模型仅考虑风险控制的不足,既能够满足不同投资者的投资偏好,又达到了“控制风险+追求收益”的双重目的。本文以油轮运价指数为研究对象,选取VLCC型、阿芙拉型以及苏伊士型每种油轮船型从2005年5月-2021年4月中188个月度数据,分别针对船型和航线的进行投资组合优化实证研究,最终得到投资组合比例的投资风险值较低且预期收益率良好。通过对不同优化目标进行灵敏度分析,可以得到:尺度和风险价值都可以度量投资风险,使用两个目标共同约束投资风险时,投资风险将得到进一步控制;熵的约束能够有效降低投资金集中度,改变预期收益,但也会影响投资风险;偏度指数对预期收益率的影响很大,比较适合激进偏好的投资者。此外,本文基于不同目标偏好建立了对比模型,结果表明:通过增加相应目标约束的偏好值可以有效改善结果中相应目标的数值;香农熵最初对收益率的影响要大于基尼指数,但是随着偏好度的上升,香农熵对模型的影响力度逐渐低于基尼指数对于模型的影响力;尺度参数在偏好值为5内对模型影响程度最大,模型实际运用时需要设立合理的偏好度;单项偏好值过高之后其他目标约束会失去影响,使模型变成单目标模型,故目标约束并不能随偏好值增加而持续增加,拥有自身的上界或者下界。更多还原

【Abstract】 The international shipping industry is a capital-intensive service industry with high investment risks.In recent years,the oil tanker market has continued to fluctuate in freight rates,which has greatly affected the operating income of tanker transportation companies.Tanker investment has become the most important business activity of tanker transportation companies.One of the contents has a strong strategic significance.The tanker market has the characteristics of large investment amount,long payback period and high risk,which determines the complexity and uncertainty of investment.Therefore,reasonable investment allocation will not only enhance the company’s investment income,but also enhance the enterprise’s anti risk ability,expand the scale of the company’s fleet and improve the enterprise’s competitiveness.Therefore,the optimization of investment portfolio allocation is very important.This paper uses a stable distribution that is more consistent with the actual distribution to characterize the skewness,excessive kurtosis,and thick tail of the oil tanker’s yield.The article measures risk through scale and value at risk Va R;increases the probability of obtaining excess returns through skew index;reduces the impact of historical data through entropy,and avoids excessive concentration of investment funds;through preference constraints,it can effectively satisfy different investors Investment needs.Combining the value at risk,scale parameter,skew index and entropy,a stable distribution-based“mean-Va R-scale parameter-skew index-entropy” multi-objective tanker portfolio optimization model is established,and the preference for target optimization is added The setting improves the traditional investment portfolio model,which only considers the shortcomings of risk control,which not only meets the investment preferences of different investors,but also achieves the dual purpose of “controlling risks + pursuing returns”.This paper takes the oil tanker freight index as the research object and selects the monthly data of the three oil tanker types of VLCC,Aframax and Suez for the fifteen years from May 2005 to April 2021,respectively,focusing on the ship type investment portfolio and route investment.The empirical research on the portfolio is conducted,and the final investment portfolio ratio has a low investment risk value and a good expected rate of return.Afterwards,through sensitivity analysis of different optimization objectives,it can be obtained that both scale and value at risk can measure investment risk.When two objectives are used to constrain investment risk together,investment risk will be further controlled;entropy constraint can effectively reduce the concentration of investment funds,Change the expected return,but also affect the investment risk;the skewness index has a great influence on the expected return rate,which is more suitable for investors with aggressive preferences.At the same time,this paper establishes a comparative model based on different target preferences,and the results show that the value of the corresponding target in the result can be effectively improved by increasing the preference value of the corresponding target constraint;Shannon entropy initially has a greater impact on the rate of return than the Gini index,but with preference The influence of Shannon entropy on the model is gradually lower than the influence of Gini index on the model;the scale parameter has the greatest influence on the model when the preference value is 5,and a reasonable preference degree needs to be established when the model is actually used;single preference value If it is too high,other target constraints will lose their influence and make the model become a single target model.Therefore,the target constraints cannot continue to increase as the preference value increases,and they have their own upper or lower bounds.更多还原