Machine learning is well-placed for reserving

Published in: Risk, Risk management, Corporate strategy, Software - IT

Companies: Old Mutual Insure

Applying machine learning (ML) technology to the reserving process is proving “interesting and fruitful” and can address longstanding issues, according to South Africa’s Old Mutual Insure.  

Ronald Richman, chief actuary at Old Mutual Insure, said there is currently a “myriad of techniques and subjective choices” in the reserving process that can be targeted with ML.

“Sometimes there is very little quantitative basis for picking a particular number,” said Richman. 

“With ML we can put this on a more objective footing”.

Richman added there is scope for further ML deployments in actuarial activities.

“We are working on a project on yield curve forecasting, together with interest rate risk. We’ve got some very nice initial results there.

“Even though the work has not been published yet, that might be one of the first actuarial applications.”

He also forecast ML models could produce a more accurate view of valuations of guarantees and options in life insurance products, including annuities, currently described by Richman as a “heavy duty tasks” for actuaries.

The full analysis is included in The Future of Risk Modelling supplement, a jointly produced report by InsuranceERM and FIS.

Click HERE to download your free PDF copy.

Paul Walsh