Leveraging machine learning and SHELDUS data to discern daily impact of disaster losses on mortgage default investment

Leveraging machine learning and SHELDUS data to discern daily impact of disaster losses on mortgage default investment Leveraging machine learning and SHELDUS data to discern daily impact of disaster losses on mortgage default investment Insurance companies are among the largest groups of intermediaries in the financial market, holding more than $33 trillion in assets around the globe. Their investment activities play a critical role in supporting liabilities and mitigating potential underwriting losses for an insurer. To manage the risk of insolvency, insurance regulators impose a set of strict rules on the asset classes in which insurance companies can invest. Typically, financial instruments held in insurance companies' asset portfolios include bonds, stocks, mortgage-related holdings, and policy loans. It has been recognized that natural disasters are occurring with increasing frequency and severity, due to changing climate. They can induce significantly negative impacts on the economy. That is why numerous studies had been devoted to studying the relationship between disaster occurrences and the performance of stocks and interest rate-related investments (Blau et al., 2008, JRI ; Lamb, 1995, JRI ; Rhomann, 2013, JRI ). However, what remains unexploredto the best of our knowledgeis the impact analysis of natural disasters on mortgage-related holdings which represent a notable portion of insurers investment portfolio.