High water: flood predictor model for insurers

Predicting Floods and Claims for Insurers using Hierarchical Spatial Regression

Using rainfall and temperature data from weather stations across Australia, a team of actuaries from the Department of Economics are developing a model to help insurers assess the severity and frequency of flood claims.

Traditionally, flood prediction involves input from insurers, civil engineers, meteorologists, research scientists and academics, with a 'bottom-up approach' undertaken to assess water levels at various times during the rainfall and weather cycle.

But  the Centre for Actuarial Studies' Kevin Fergusson, Enrique Calderin and Xueyuan Wu are using a Faculty of Business and Economics research grant to investigate flood occurrence using a 'top-down approach' to create the Extreme Rainfall Event (and associated Flood Event) Model.

As part of their research, the trio will examine climatic data, including rainfall and temperature data at weather stations across Australia, various climatic indices such as the Indian Ocean Dipole, Madden-Julian Oscillation Index, Southern Annular Mode Index and Southern Oscillation Index, as well as solar and lunar quantities such as sunspot numbers and lunar apogees and perigees.

Once completed, their model will feed into the overarching model used to predict claims and set appropriate solvency levels for insurers.


The spatial methodology used for modelling and applying flood risks in Australia will create a useful tool for insurers in Australia and across the world.

The model will also

  • improve the prediction accuracy of flood events,
  • provide better estimates of required solvency levels of insurers,
  • potentially enhance return on capital of insurers, and
  • build relationships with the insurance industry in Australia and overseas.


K. Fergusson, 'Preliminary Work on the Prediction of Extreme Rainfall Events and Flood Events in Australia', presented at the Australian Institute of Actuaries General Insurance Seminar, Melbourne, November 2016.