If the past two years have taught us anything, it’s that human behavior doesn’t always conform to mathematical formulae and projections, especially in complex and dynamic environments like global finance. Consequently, as companies move towards risk-based solvency capital more accurate modeling of the extreme behavior of the enterprise can translate into real capital benefits.
In “Keeping it Simple” (login required), appearing in InsuranceERM, London-based Milliman actuary Neil Cantle makes a convincing case for simple agent-based approaches for modeling complex financial behaviors.
He finds that because “model complexity and run-time are in constant tension,” behavior-based approaches that aim to realistically model the features of business portfolios also tend to increase calculation time, mainly because of their complexity.
A solution may lie in agent-based models, which can model complex behaviors by using very simple rules.
As Cantle points out, the concern to account for all factors and forces may actually produce a paradoxical effect.
“There is possibly an irony in the fact,” Cantle concludes, “that the future direction of modelling into behaviour-based approaches . . . will actually require us to make the structure of the models simpler in order to study more complex behaviors!”