“It's far too early to say that better models would lead to better products,” but that's the hope, Mr. Hodgson said. He needs to prove the concept will work before getting internal funding to build the model.
While models based on modern portfolio theory assume all agents will act the same way, agent-based models assume a world of complex interactions among agents. Agents can be people, types of investors (e.g., hedge funds or passive equity funds) or policymakers such as the Federal Reserve Board.
The idea to build an agent-based model came from the Santa Fe Institute, which began a project in 2009 to model the U.S. economy. The project by the independent research center, which focuses on understanding complex adaptive systems, is ongoing.
At Towers Watson, such a model could help predict investment environments, or design new pension funds, Mr. Hodgson said, emphasizing it's too early to say what uses the model might eventually have.