Tokio Millennium Re (TMR) is an international reinsurance company which covers the large pay-outs insurance providers have to make in the wake of catastrophic weather events.
Accurately projecting losses from events such as typhoons is critical to the company, but it is also a highly complex and difficult task, full of uncertainties.
In China, one or two major typhoons can cause damage to property and business which amounts to millions of dollars. TMR is committed to a multi-model approach and license all models available in the region. Despite this, lack of detailed data on past claims made by insurance companies makes it particularly difficult to predict their maximum annual loss for China.
Our team of mathematicians from the Institute of Mathematical Innovation (IMI) worked with TMR to better understand the historical insurance claim data from two regions in China, Fujian and Guangdong. Located on the coast, these regions are home to a number of major business parks, and are regularly affected by typhoons.
Working alongside academic staff from the Department of Mathematical Sciences and the Department of Architecture and Civil Engineering at the University of Bath, we looked at reinsurance claim information from the previous 10 years. Using advanced statistical techniques, we used this weather pattern related data to more precisely inform the models for predicting losses.
Enabling improved insurance fund management
TMR used our analysis to adjust their simulation model output across their entire portfolio in China.
By enabling TMR to create more accurate insurance loss projections, our work has also allowed the company to improve their insurance funds management.
Forecasting insurance losses seasons in advance
Following the success of the first project, we were re-commissioned a year later to review new data and to create a new prototype simulation model.
Bringing together records from all typhoons that have hit the region since the 1950s, rather than just the figures from typhoons that resulted in claims, as well as additional environmental variables, our team were able to create a new and more accurate prototype loss simulation model based on real-world data.
Having access to more accurate predictions of losses means that TMR could potentially forecast losses seasons in advance. This has given the company an even more accurate understanding of the reinsurance risk they are exposed to, allowing them to further improve their management of insurance funds.