The Study Of Operational Risk Measurement In Insurance Companies

In resent years, operational risk has become one of the focuses in international financial fields. However, the researches of operational risk in insurance companies are comparatively less than in banks. Risk measurement is the foundation of the risk management. Therefore the risk system, the traits of operational risk in insurance companies and its measurement approach are researched in this dissertation.The definition and the traits of operational risk are introduced and the relevant research fruit are summarized. The influence of the new Basle to the operational risk management in insurance companies is discussed. The modern risk system of insurance company is built. It is scientific and consistent with the risk system in the new Basel which represents the risk system of modern financial companies. The traits of operational risk and operational risk management status quo in the insurance companies in China are analyzed. The operational risk causes are categorized into four levels.According to the shortages of general operational risk data model, a new model-operational topological data model was brought forward. In this model, a risk loss event is attributed to risk trigger cause which lead to the initial loss and successor causes which affect the initial loss. Operational risk topological data model is composed of the graph and node data. The graph pictures how each risk cause leads to the loss event. The nodes data includes the initial loss and the loss effect multiplier of successor cause to its previous causes. Six basic principles of recording operational risk events with topological data model were brought forward.On the basis of operational risk topological data model, the influence diagrams were applied in the operational risk measurement. The operational risk influence diagram takes all the possible trigger causes as the initial nodes, takes their successor cause as the chance nodes and the 'operational risk event' as the final node. It includes the possible paths which represent all kinds of operational risk events in the company. Each record of operational risk event by the topological data model is a sample of a certain path in the operational risk influence diagram of the company. The operational risk influence diagrams with control/process type of cause nodes and without control/process type of cause nodes were constructed separately, and they can be combined directly into one complete operational risk influence diagram if needs. The basic guidelines which the influence diagrams construction needs were brought forwardThe three steps of calculating operational risk influence diagrams based on the topological data model is that calculating the loss severity distributions of all paths and sum them up to get the company' s operational risk loss severity distribution by the. influence diagrams probabilistic inference; calculating the company' s operational risk loss frequency distribution by the frequency of trigger causes; and combining them into an aggregated operational risk loss distribution. The core operation processes of the insurance company were separated out to demonstrate the calculation steps of the operational risk influence diagram with the subjective data got by a survey.To make the employee report the operational risk event happened already in stead of adopting unauthorized or deregulation operations to hide or remedy, a dual-purpose incentive mechanism model for operational mistakes control was constructed and it was analyzed with the game theory and the principal-agent theory. The theoretical Pareto dominance of the model can not be implemented in practice. However, the second best executable salary policy exists. Diverse salary policies need to be made according to various mistakes, which with different possible loss, hardness of audit and increase in loss caused by delay disposition.Influence diagrams approach based on topological data model is a new effective method of integrated analyses and assessment of operational risk. It focus on the failures of control/process, makes the use of the information contained in the history data more sufficiently, simplifies the analysis of correlation between all the causes and can be modified and improves agilely according to new information, improve the suitability of external data and support the operational risk management decision.