The paper motivates and describes a regime-switching macro-driven simulation model for the purposes of simulating long horizon asset returns. The paths generated by this model are compared to more common approaches – multivariate normal generators and a block bootstrap simulation. Despite calibration to the same mean and variance in returns, the models display divergent behavior in the tails of long horizon return simulations. Simulations are run through representative defined contribution and defined benefit applications to examine the filtered behavior and draw inferences for future applied research and application.