Stochastic Asset And Liability Management Of Insurers Based On Multi-objective Programming

Asset and liability management (ALM) is the core of insurance company’s management. Because of the limitations of technology, most of or raditional ALM technologies are single objective models. The management and operation in insurance company has multi-objective characteristic. It has many goals, which are complementary, contradictory or independent to others. Single objective models only pay attention to asset management or liability management, which cannot get the optimal solution under multi objectives and cannot achieve the asset and liability function at the higher level. Multi-objective programming can solve these problems. Besides, determinate multi-objective ALM models’predictive ability is poor. Stochastic ALM models can make effective decision under stochastic economic conditions and uncertain liability’s fluctuation. If we can put the multi goals and stochastic characters into decision process, the ability of insurance company against risks will be significantly enhanced.The main contents are as follows:1. Literature review of asset liability management and multi-objective programming. I sorted out theory from traditional ALM technology, ALM on stochastic programming and ALM on optimal stochastic control. I also reviewed the multi-objective programming from theory research, applied research and solving research.2、Providing a multi-objective ALM decision-making systems. First, I explained the multi-objective attribution of insurance company and gave multi-objective decision process of asset allocation, capital management and liability structure. Next, an insurance ALM framework was developed, which was different from other financial institution in asset and liability characteristics. Finally, I established multi-objective ALM decision-making system and focused on the selection of company’s objectives and constraints.3.Summarizing the existing scenarios generation methods, comparing them, and generating asset and liability scenarios. First, I used dynamic Nelson-Siegel term structure model and vector auto-regression model to generate asset return scenarios. Then the premium, payment, actuarial reserve and dividend scenarios were generated based on stochastic3、Random balance scenario generation, collation of existing scenario generation method and compare the advantages and disadvantages of each method. Dynamic Nelson-Siegel term structure models and vector autoregression model to generate capital gains rate scenario, given the number of random changes in the policy equation, mortality, lapse rates, dividend policy, and has been the traditional non-participating insurers on the basis of participating insurance premiums, claims, actuarial reserves, bonus scenario generation models. Empirical research assets and liabilities based on randomly generated scenarios of macroeconomic data and experience in the insurance industry data.4、Life insurance companies multi-objective asset and liability management model. The importance of the choice of target profit targets based on the value of the target, the target risk, solvency constraints and changes in the value of assets and liabilities constraints as objectives and constraints of life insurance model. In solving multi-objective planning, the use of genetic algorithm, first summarizes the main steps for solving ideas and were given the traditional non-participating insurance, banking-type insurance dividends, asset-liability protection-type insurance dividends decision results and the company’s overall level. Mortality, lapse rates, scheduled interest expense ratio several key assumptions sensitivity analysis of life insurance. According to the Chinese insurance market, the insurance company’s capital structure, asset size, business structure and other indicators of the insurance company is divided into different categories, select the business objectives of different periods and in accordance with the characteristics of the insurance companies to quantify, the above model appropriate amendments, optimal results were analyzed according to the model given.5、Insurance companies stochastic asset-liability management. Property and casualty insurance industry is different from the life insurance industry is mainly characterized by short-term highly liquid assets, liabilities and insurance, construction and insurance ALM therefore different asset and liability management model in the model objective function and constraints select areas. In this part of the research is still conducted a sensitivity analysis and the application of financial insurance asset and liability management.The main innovation of the paper is as follows:1、Consider the assets and liabilities of the double random factors proposed debt scenario generation method. Generally, asset scenario generation method is more mature and fixed, and insurance products include many policy options as well as various regulatory restrictions, resulting in changes in the liability insurance company has great complexity. In this paper, the model analysis point, given the debt scenario generation method. At the same time, in the past due to the calculation of technical constraints, asset-liability management study only assume that most of these assets yield a random factor, the biggest difference is that the insurance company’s assets and liabilities management randomness of liability, so this article add liabilities random scenario, then get random assets liability management model.2、Multi-objective asset and liability management model to break the existing asset and liability management target mode. Due to the limitations of technology, the target previously set by the asset and liability management objectives only concerned liability management or asset management can be achieved at this level, the paper studies the existing asset-liability management of insurance companies to extend and enhance the target, and quantify, insurance companies realize the asset and liability management function from a higher level.3、Using intelligent algorithm for solving multi-objective stochastic asset and liability management model. From the multi-objective theory, problem solving multi-objective model is one of the most critical issues, and the results of multi-objective planning theory should be non-inferior solution set, but due to difficulties in solving large, it is generally converted into a single multi-objective goal, but in the conversion process requires adding a variety of conditions, so it has a lot of subjectivity and limitations, so the result is not general. Pareto will solve the problem, you can make the results to a more wide range. Taking into account the decision-makers understand the preference in case of need to get a unique solution, the study of this thesis for Pareto optimal solution and have carried out a unique study.4、Liability management model and its application to solve the result of random assets, given the investment decision-making based on the principles and guiding significance. Different insurance company’s development strategy is different, but with an insurance company at different stages of development strategy will be different. This is the first decision of the company ALM different development strategies and different development periods of empirical research, get a lot of concrete and effective asset and liability management principles to guide managers work for correct asset-liability management decisions is critical.