本文選題：多目標規(guī)劃 + 隨機情景生成�。� 參考：《南開大學》2014年博士論文

【摘要】：資產(chǎn)負債管理(ALM)是金融機構(gòu)經(jīng)營管理的核心內(nèi)容,其早期作為重要的風險管理手段,能夠管理利率風險、流動性風險或匯率風險,近年來資產(chǎn)負債管理的內(nèi)容逐漸擴展至資本管理、利潤規(guī)劃與發(fā)展等方面,成為金融機構(gòu)重要的管理手段。然而,由于技術(shù)的局限性,傳統(tǒng)的資產(chǎn)負債管理大多是基于單目標的最優(yōu)決策模型,事實上,保險公司經(jīng)營具有多目標屬性,其要實現(xiàn)的目標有多個,單目標資產(chǎn)負債管理技術(shù)無法滿足保險公司綜合性管理需求,無法得到多種目標下的最優(yōu)決策結(jié)果,多目標最優(yōu)決策模型則可以解決這一問題。此外,確定性管理模型不具備良好的預測能力,而隨機管理模型能夠根據(jù)宏觀經(jīng)濟形勢、資本市場隨機變化、公司自身資產(chǎn)負債價值波動等特征做出有效決策。如果能夠?qū)⒍囗椖繕�、資產(chǎn)與負債隨機變化等特征同時納入到?jīng)Q策過程中,便可以使保險公司得到更加全面、科學的資產(chǎn)負債管理決策,顯著提高其綜合管理能力。 論文的主要內(nèi)容如下： 1、金融機構(gòu)資產(chǎn)負債管理與多目標規(guī)劃理論研究與文獻綜述,梳理了資產(chǎn)負債管理理論與多目標規(guī)劃理論,分別從傳統(tǒng)ALM技術(shù)、基于隨機規(guī)劃的ALM和基于隨機控制的ALM三個方面進行了分析和總結(jié),并對多目標規(guī)劃理論研究、應用研究和求解研究三方面進行了文獻整理和綜述。 2、提供了一個完整通用的保險公司多目標資產(chǎn)負債管理決策系統(tǒng)。首先闡述了保險公司經(jīng)營的多目標屬性,給出資產(chǎn)配置、資本管理、險種結(jié)構(gòu)三種決策的多目標決策過程。其次構(gòu)建了保險公司資產(chǎn)負債管理框架,指出保險公司資產(chǎn)負債管理與其他金融機構(gòu)的最大不同之處在于資產(chǎn)和負債的特性,并對其進行分析。最后建立了保險公司多目標資產(chǎn)負債管理決策系統(tǒng),并重點討論了公司的目標函數(shù)選擇和約束條件。 3、隨機資產(chǎn)負債情景生成,提出壽險負債情景生成方法。整理現(xiàn)有的一般化情景生成方法并比較各種方法的優(yōu)劣。利用動態(tài)Nelson-Siegel利率期限結(jié)構(gòu)模型和向量自回歸模型生成資產(chǎn)收益率情景,給出了保單數(shù)量、死亡率、退保率、分紅政策的隨機變化方程,并在此基礎上得到壽險保費、賠款、精算準備金、紅利情景生成模型。根據(jù)我國宏觀經(jīng)濟數(shù)據(jù)以及保險業(yè)經(jīng)驗數(shù)據(jù)進行資產(chǎn)與負債隨機情景生成的實證研究。 4、壽險公司多目標資產(chǎn)負債管理模型。根據(jù)目標的重要程度選擇利潤目標、價值目標、風險目標、償付能力約束與資產(chǎn)負債價值變化約束作為壽險模型的目標和約束。在多目標規(guī)劃求解方面,采用遺傳算法求解,分別給出了傳統(tǒng)不分紅保險、理財型分紅險、保障型分紅險和公司整體層面的資產(chǎn)負債決策結(jié)果。在各險種的敏感性分析方面,重點討論死亡率、退保率、預定利率、費用率等壽險產(chǎn)品定價關(guān)鍵假設,在公司層面敏感性分析上,則主要討論發(fā)展戰(zhàn)略、資本和規(guī)模增長率的對決策結(jié)果的影響。根據(jù)我國保險市場上壽險公司的資產(chǎn)規(guī)模、盈利能力、資本結(jié)構(gòu)、成長性等指標將保險公司分為不同的類別,根據(jù)保險公司的特點以及不同時期的經(jīng)營目標的偏好,得到最優(yōu)決策結(jié)果并進行分析。 5、財險公司隨機資產(chǎn)負債管理。財產(chǎn)保險行業(yè)不同于壽險業(yè)的主要特點在于負債的短期性和保險資產(chǎn)的高流動性,因此目標函數(shù)與約束條件選擇了利潤目標、風險目標、流動性約束、償付能力約束以及資產(chǎn)價值變化約束。相比壽險險種,非壽險產(chǎn)品的險種差異較小,因此主要給出公司層面的資產(chǎn)負債管理決策結(jié)果。在敏感性分析中主要討論了賠付率、資本和戰(zhàn)略目標對決策結(jié)果的影響。 論文的主要創(chuàng)新點如下： 1、考慮資產(chǎn)與負債雙重隨機因素,提出了負債情景生成方法。一般情況下資產(chǎn)情景生成的方法較為成熟,而保險產(chǎn)品包含很多保單選擇權(quán)及監(jiān)管制約,導致負債變化具有極大的復雜性。本文在給出保單數(shù)量、死亡率、退保率、分紅政策的隨機變化方程基礎上根據(jù)保險精算原理得到保費、賠款、精算準備金、紅利等負債情景,建立雙隨機資產(chǎn)負債管理模型。 2、多目標資產(chǎn)負債管理模型克服了單目標資產(chǎn)負債管理的局限性。以往的資產(chǎn)負債管理往往僅追求盈利或風險管理的單一層次管理,并且設定的目標只關(guān)注本層級能夠?qū)崿F(xiàn)的目標,范圍較窄,本論文將這些目標進行擴展及提升,從更高層次實現(xiàn)保險公司資產(chǎn)負債管理的功能。 3、利用智能算法求解隨機多目標資產(chǎn)負債管理模型。從多目標理論而言,多目標模型的求解問題是最關(guān)鍵的問題之一,而理論上多目標規(guī)劃的結(jié)果應該是非劣解集,但由于求解困難較大,所以一般將多目標轉(zhuǎn)換為單目標,但在轉(zhuǎn)換過程中需要各種條件的加入,因此具有很大的主觀性和局限性,使結(jié)果不具有一般性。非劣解集則解決了該問題,可以使結(jié)果應用于更加寬的范圍�？紤]到?jīng)Q策者在了解偏好的情況下需要得到唯一解,所以本論文的研究中對于非劣解集和唯一最優(yōu)解都進行了研究。 4、隨機資產(chǎn)負債管理模型及其求解結(jié)果的應用,給出了具有指導意義的投資原則和決策依據(jù)。不同的保險公司發(fā)展戰(zhàn)略不同,而同一家保險公司在不同的發(fā)展階段戰(zhàn)略目標也會不同。本文首次對不同發(fā)展戰(zhàn)略和不同發(fā)展時期的公司ALM決策進行實證研究,得到許多具體而有效的資產(chǎn)負債管理建議,這些建議對于指導管理者得到正確全面的資產(chǎn)負債管理決策極為重要。
[Abstract]:Asset Liability Management (ALM) is the core content of the management of financial institutions. Early as an important means of risk management, it can manage interest rate risk, liquidity risk or exchange rate risk. In recent years, the content of asset liability management has gradually expanded to capital management, profit planning and development and so on, which has become an important management means for financial institutions. However, due to the limitation of technology, the traditional asset liability management is mostly based on the optimal decision model based on single objective. In fact, the insurance company has multiple objectives, and it has many goals to achieve. The single objective asset liability management technology can not meet the comprehensive management needs of the insurance company and can not get the most objective. The multi objective optimal decision model can solve this problem. In addition, the deterministic management model does not have good forecasting ability, and the stochastic management model can make effective decisions based on the macroeconomic situation, the random change of capital market, the value fluctuation of the company's own assets and liabilities, and so on. The random changes of assets and liabilities are incorporated into the decision-making process, and the insurance companies can get more comprehensive and scientific assets and liabilities management decisions and significantly improve their comprehensive management ability.
The main contents of the paper are as follows:
1, the theoretical research and literature review of assets and liabilities management and multi-objective programming in financial institutions, combing the theory of asset liability management and multi-objective programming, analyzing and summarizing the three aspects of the traditional ALM technology, ALM based on random programming and ALM based on random control, and the research on the theory of multi-objective programming, application and research. Three aspects of solution research are reviewed and summarized.
2, it provides a comprehensive and universal multi-objective asset liability management decision system for insurance companies. First, the multi-objective attributes of the insurance company are expounded, and the multi-objective decision-making process of the assets allocation, capital management and the insurance structure of three kinds of decision-making is given. Secondly, the management framework of the insurance company's assets liabilities is constructed, and the assets and liabilities of the insurance companies are pointed out. The biggest difference between management and other financial institutions lies in the characteristics of assets and liabilities, and analyzes them. Finally, the multi target asset liability management decision system of insurance companies is established, and the selection and constraints of the objective function of the company are emphatically discussed.
3, the situation of the random asset liability scenario is generated, and the generation method of the life insurance liability scenario is put forward. The existing general scenario generation method is arranged and the advantages and disadvantages of various methods are compared. The dynamic Nelson-Siegel interest rate term structure model and the vector autoregressive model are used to generate the asset return scenario, which gives the policy number, death rate, reinsurance rate, and dividend policy. On the basis of this, we get the life insurance premiums, indemnities, actuarial reserves, and the dividend scenario generation model. The empirical research on the formation of the random situation of assets and liabilities based on the macroeconomic data and the insurance industry experience data in China.
4, the multi-objective asset liability management model of life insurance company. According to the importance of the target, we choose the profit target, the value target, the risk target, the solvency constraint and the asset and liability value change constraint as the goal and constraint of the life insurance model. In the multi objective programming solution, the traditional non dividend insurance is given by using the genetic algorithm. In the sensitivity analysis of various types of insurance, we focus on the key assumptions of the pricing of life insurance products such as death rate, reinsurance rate, predetermined interest rate, cost rate and so on. In the sensitivity analysis of the company level, the development strategy, capital and scale growth are mainly discussed. According to the assets scale, profitability, capital structure, growth and other indicators of life insurance companies in China's insurance market, the insurance companies are divided into different categories. According to the characteristics of the insurance companies and the preference of the operating targets in different periods, the optimal decision results are obtained and analyzed.
5, the property insurance company's random asset liability management. The main characteristics of the property insurance industry different from the life insurance industry are the short-term liabilities and the high liquidity of the insurance assets. Therefore, the target function and the constraint conditions choose the profit target, the risk target, the liquidity constraint, the solvency contract and the change of the asset value. There is a small difference in the risk of non life insurance products. Therefore, the results of asset liability management decision are mainly given at the company level. In the sensitivity analysis, the impact of the loss rate, capital and strategic objectives on the decision results are mainly discussed.
The main innovations of the paper are as follows:
1, considering the dual random factors of assets and liabilities, a method of generating debt situation is proposed. Under the general situation, the method of asset scenario generation is more mature, and the insurance product contains a lot of policy options and regulatory constraints, which leads to the great complexity of the debt change. In this paper, the number of policies, the mortality rate, the rate of reinsurance and the dividend policy are given in this paper. On the basis of the machine change equation, according to the actuarial principle, we get the debt situation of premium, reparations, actuarial reserves, dividends and so on, and establish a double stochastic asset liability management model.
2, the multi objective asset liability management model overcomes the limitation of the single objective asset liability management. The past assets and liabilities management often only pursue the single level management of profit or risk management, and the goal is only concerned with the goal that can be realized at the level of this level. The functions of assets and liabilities management of insurance companies are realized at different levels.
3, using the intelligent algorithm to solve the stochastic multi-objective asset liability management model. From the multi-objective theory, the solving problem of the multi-objective model is one of the most critical problems, and the result of the multi-objective programming in theory should be a non inferior solution set, but because of the difficulty of solving the problem, the multiobjective is converted into a single target, but in the process of conversion, the multi-objective model can be converted into a single target. It requires a variety of conditions to join, so it has a lot of subjectivity and limitations, so that the result is not general. The set of non inferiority solutions can solve the problem and can apply the result to a wider range. Considering that the decision-maker needs a unique solution in the case of understanding the preference, the study in this paper is not a bad solution set and the only one. The optimal solution is studied.
4, the application of the stochastic asset liability management model and its solution results give the guiding significance of the investment principles and decision-making basis. Different insurance companies have different development strategies, while the same insurance company will have different strategic goals at different stages of development. This paper is the first of the different development strategy and the different development period of the company ALM. A number of specific and effective asset liability management recommendations are obtained by empirical research, which are very important to guide managers to get a correct and comprehensive asset liability management decision.
【學位授予單位】：南開大學
【學位級別】：博士
【學位授予年份】：2014
【分類號】：F842.3;F840.4

【相似文獻】

相關(guān)期刊論文 前10條

1 李薛青,鄭澤華;試析我國保險業(yè)資產(chǎn)負債管理模式的選擇及體系構(gòu)建[J];預測;2005年05期

2 張?zhí)煜?張宗成;;淺談資產(chǎn)負債管理策略[J];當代經(jīng)濟;2007年04期

3 張晶;;我國保險業(yè)資產(chǎn)負債管理探析[J];生產(chǎn)力研究;2008年23期

4 岳萍娜;李哲;;資產(chǎn)負債管理理論述評[J];區(qū)域金融研究;2009年11期

5 黃劍;李石凱;;日本銀行業(yè)資產(chǎn)負債管理的歷史演進及啟示[J];現(xiàn)代日本經(jīng)濟;2012年02期

6 范欽建;論銀行資產(chǎn)負債管理[J];吉林財貿(mào)學院學報;1990年05期

7 ;實行資產(chǎn)負債管理 建立自我規(guī)范經(jīng)營機制[J];中國金融;1991年08期

8 張曉暉;;資產(chǎn)負債管理模式簡介[J];天津金融月刊;1993年05期

9 ;商行資產(chǎn)負債管理辦法出臺[J];技術(shù)經(jīng)濟信息;1994年09期

10 張紹瑞;我國銀行資產(chǎn)負債管理存在問題及對策[J];技術(shù)經(jīng)濟與管理研究;1994年05期

相關(guān)會議論文 前3條

1 李祝用;王慶松;;保險金融集團資產(chǎn)負債的匹配管理——保險業(yè)綜合經(jīng)營形勢下的風險管控[A];十二五·新挑戰(zhàn)：經(jīng)濟社會綜合風險管理——北大賽瑟（CCISSR）論壇文集·2011[C];2011年

2 王福新;易丹輝;;未定權(quán)益模型在資產(chǎn)負債管理中的慶用[A];2002年中國管理科學學術(shù)會議論文集[C];2002年

3 史鵬;柏滿迎;;一個基于隨機控制方法的養(yǎng)老基金管理模型[A];2004年中國管理科學學術(shù)會議論文集[C];2004年

相關(guān)重要報紙文章 前10條

1 包洪波 楊紅宇;資產(chǎn)負債管理工作的探討[N];中國城鄉(xiāng)金融報;2004年

2 本報記者 鄭曉敏;提高資產(chǎn)負債管理水平[N];中國城鄉(xiāng)金融報;2001年

3 ;中國農(nóng)業(yè)銀行資產(chǎn)負債管理先進個人名單[N];中國城鄉(xiāng)金融報;2004年

4 伊編;農(nóng)行表彰資產(chǎn)負債管理先進[N];中國城鄉(xiāng)金融報;2004年

5 林薇;推動資產(chǎn)負債管理工作再上新臺階[N];中國城鄉(xiāng)金融報;2004年

6 記者 胡萬年 徐亞麗;全面推進資產(chǎn)負債管理[N];金融時報;2001年

7 本報記者 羅進;農(nóng)行啟動資產(chǎn)負債管理信息系統(tǒng)測試[N];中國城鄉(xiāng)金融報;2008年

8 記者 陳玉潔;中國電財創(chuàng)新資產(chǎn)負債管理見成效[N];國家電網(wǎng)報;2010年

9 申夢玉　段辰菊;加強資產(chǎn)負債管理 服務保險發(fā)展方式轉(zhuǎn)變[N];中國保險報;2010年

10 記者 周萃;貸款增長加快 資產(chǎn)負債管理待優(yōu)化[N];金融時報;2012年

相關(guān)博士學位論文 前7條

1 趙偉;金融機構(gòu)資產(chǎn)負債管理模型和運用的新發(fā)展[D];武漢大學;2005年

2 金秀;資產(chǎn)負債管理多階段模型及應用研究[D];東北大學;2006年

3 解強;基于多目標規(guī)劃的保險公司資產(chǎn)負債管理[D];南開大學;2010年

4 馮鵬熙;我國商業(yè)銀行資產(chǎn)負債管理的實證研究[D];華中科技大學;2006年

5 王建偉;基于控制論的保險公司資產(chǎn)負債管理研究[D];湖南大學;2006年

6 田君;基于資產(chǎn)負債管理的壽險投資問題研究[D];同濟大學;2006年

7 薛峰;商業(yè)銀行經(jīng)營管理模式的應用研究[D];大連理工大學;2000年

相關(guān)碩士學位論文 前10條

1 陳春曄;財務公司資產(chǎn)負債管理研究[D];首都經(jīng)濟貿(mào)易大學;2008年

2 王抒敏;壽險業(yè)資產(chǎn)負債管理研究[D];吉林大學;2006年

3 蘇慧;我國壽險公司資產(chǎn)負債管理研究[D];東北財經(jīng)大學;2007年

4 梁鑫鑫;資產(chǎn)負債管理在哈爾濱銀行的應用研究[D];黑龍江大學;2010年

5 胡蓉;資產(chǎn)負債管理在我國壽險業(yè)經(jīng)營中的運用研究[D];廣西大學;2007年

6 周伍;國有獨資商業(yè)銀行資產(chǎn)負債管理研究[D];安徽大學;2004年

7 王爽;基于多目標規(guī)劃的保險公司資產(chǎn)負債管理定量方法的研究[D];重慶工商大學;2012年

8 呂成佳;我國壽險公司隨機資產(chǎn)負債管理建模與分析[D];西南財經(jīng)大學;2008年

9 張佳昌;金融機構(gòu)資產(chǎn)負債管理模型及在泉州銀行的應用[D];華僑大學;2011年

10 趙培娟;養(yǎng)老基金動態(tài)資產(chǎn)負債管理模型與分析[D];大連理工大學;2009年

，

本文編號：2018200