Research on the Sustainability of Property Insurance under the Background of Extreme Weather Events

Abstract

In recent years, the frequency of extreme weather events has significantly increased, resulting in global losses exceeding $1 trillion. The profit crisis of insurance companies and the issue of homeowners' payment ability have become more apparent. This article focuses on the sustainability of property insurance under extreme weather events, establishes a comprehensive insurance and protection model, and provides recommendations for relevant departments.Firstly, a set of seven indicators including the proportion of household insurance expenditure was selected to establish an indicator system. After conducting Spearman correlation coefficient testing on the data and combining it with LSTM neural network to predict the frequency of extreme weather in the next decade, this paper uses the SVM binary classification model to classify high-risk areas as uninsurable areas and other areas as insurable areas. This decision-making process resulted in insurance companies covering areas such as Alabama and refusing to underwrite insurance in Arizona, while also making relevant recommendations.Secondly, by calculating indicators for communities and real estate developers, and considering the impact of population growth on housing demand, this article uses the ARIMA model to predict population growth in different regions over the next five years. After standardizing the data, the entropy weighting method is applied to consider the subjective importance of different real estate developers for different indicators and assign weights to the indicators. Weighted indicator data is used as input for the SVM model, dividing regions into two categories, with regions with better overall conditions being more favorable for real estate development. This has led people to prioritize the development of regions such as North Carolina and Indiana. In addition, this article introduces the TOPSIS regional ranking comprehensive evaluation index, which is beneficial for the development of higher ranked regions. Similar to our SVM classification results, Indiana and North Carolina rank second and third.