With global warming on its way, weather-related losses increased dramatically in the past few decades, yet there has been a large protection gap widely observed in the insurance market globally. Why cannot insurance cover weather-related losses? This research will try to tackle this disequilibrium of the flood insurance market in Japan, under the analytical framework of Evolutionary Game Theory, which to my best knowledge, has not been widely used in insurance economics. The game structure allows us to analyze the interaction (i.e., payoffs) among three main agents: property owners, insurers and the government. To make the theoretical analysis more intuitive and closer to the practical reality, Prospect Theory will be introduced into the game model. By comparing different reference points, the influence of consumers’ risk perception/appetite and the contents of insurance policies will be investigated. Then the theoretical results will be further explained through a web-based experiment and a numerical simulation. The results of this research will be discussed under institutional analysis so that some suggestions will be provided for the construction towards a ‘better’ arrangement of flood insurance not only for Japan but also for any country in need. The contents are organized in the following sequence: The first step is a review of the current market state and literature regarding economic analysis of catastrophe insurance market. In this phase, some interviews with practitioners in insurance companies and scholars in this field are necessary for a more accurate and holistic understanding. The second step is static bipartite game analysis in one period. The game model includes two participants, the property owners and insurance companies. The focus is the modelling of the market status, as well as a preliminary analysis of the factors influencing market equilibrium. The necessity of a muti-period dynamic analysis is clarified.The third step is evolutionary game analysis in multiple periods. Prospect Theory is introduced into the game model. Different reference points (no insurance, partially and fully insured; default and optional flood coverage as reference points) are tested under the game structure. The fourth step, introduce government intervention into the model. After analyzing the feasibility of intervention from the public sector. Modelling some representative public financial ways (such as public insurance pool, public-backed reinsurance etc.) of intervention into the evolutionary game model. The fifth step is focused on quantitative empirical proofs of the theoretical analysis. A web-based experiment is conducted mainly for the test of reference points in Prospect Theory. Then, the observed numeric relationship is further used for a simulation with Taylor expansion and Ito-type stochastic differential equation, to test the effect of different public interventions. In the final sixth step, the results are shared in several conferences and revised based on the received feedback. The main expected outputs would be presentations at international conferences and 2~3 papers published in top journals in the insurance sector. 

With increasing greenhouse gas emissionsand urbanization (the heat island effect), floods have been accounting for alarger part of economic losses globally. The importance of insurance privateinsurance sector in participating in flood risk management increasesaccordingly. In this research, we modify the most common and internationallyaccepted method of assessing building damage due to flooding, i.e., depth–damagecurves (DDCs) into probabilistic depth–damage curves (PDDCs). Based on the generatedPDDCs, we model the premiums of flood insurance using the Great Tokyo Area asan applied example. Finally, we also quantify the impact of greenhouse gasemission scenarios on individual and aggregate losses and premiums.The outputs of this project include: Firstly,a high-resolution (50m) flood map contains the spatial distribution of waterstages. Secondly, a function between flood depth and probability of damage overcertain thresholds, i.e., probabilistic depth–damage curves (PDDCs). Thirdly,premium rating models using PDDCs. And finally, the impact of climate change oninsurance premiums.The results were submitted to APRIA 2022 annualconference, and the further revised version will be presented at IAFICO 2023 annualconference.