Modeling aggregated returns with application to segregated fund guarantees

In guarantee valuation for a segregated fund, the simulation process can be time-consuming. When simulation calculations are based upon weekly or monthly return models, the computation can be quite lengthy for contracts that extend over decades. Simulation run time can be reduced by decreasing the number of calculations. This is accomplished through an aggregated return model. We study models for the aggregated returns when the estimated model is Lognormal, an AR(1), two-state regime switching and a Multivariate Lognormal. As an illustration of the aggregate models, we use a conditional tail expectation for valuation of a segregated funds guarantee.