Calibration of Stochastic Scenario Generators for DFA

Leading actuarial companies employ stochastic simulation models to evaluate the viability of pension plans and insurance companies over a set of projected scenarios. A critical element involves generating future scenarios. We show that the problem of calibrating a stochastic scenario system can be posed as a special optimization model, and illustrate the process by means of the Towers Perrin - Tillinghast CAP: Link system. We briefly discuss solution algorithms for the resulting non-convex problem. Areas for future research are indicated.