Allstate and IBM test quantum computing for insurance portfolio management

Allstate and IBM are exploring how quantum computing could help insurers build homeowners insurance portfolios while accounting for correlated catastrophe risks.

The joint research, published on arXiv in May, applies a quantum-classical approach to the “knapsack problem,” which involves selecting the most valuable combination of items without exceeding a set capacity. In insurance, the problem can be used to determine which policies to include in a portfolio while keeping potential losses within the insurer’s risk tolerance.

The challenge is particularly relevant to homeowners insurance because catastrophes such as wildfires, hurricanes and tornadoes can damage many insured properties simultaneously. Allstate currently runs around 100,000 simulations to estimate potential portfolio outcomes, but rare events can be difficult to model accurately.

“On the homeowner side, it really requires us to be thinking from a portfolio perspective and not just from an individual risk perspective.” – Eric Huls, chief analytics officer and chief data officer at Allstate.

The researchers used IBM Quantum Heron to generate potential portfolio combinations, followed by classical computing to repair answers that exceeded the loss budget and guide subsequent rounds. The method was tested against four classical approximation methods, with each given 30 minutes to solve a problem.

For problems containing up to 75 items, the quantum-classical and classical methods both reached the optimal answer. At larger scales, the quantum approach remained competitive and slightly outperformed the strongest classical method under tight constraints.

The workflow is not yet ready for commercial use at large scales, but Allstate said the research could help prepare it to use quantum computing as the hardware improves.