This paper addresses the ever-prominent issue of how to evaluate and forecast future longevity dynamics. Indeed, studying the evolution of mortality and/or the cost of longevity risk is a major task for both demographers and actuaries. In contrast to the usual period-based evaluation, we consider the problem of approximating the distribution of future life expectancy with a cohort-based perspective. In particular, we suggest an application of the Least-Squares Monte Carlo approach, which allows to overcome the straightforward nested simulations method. The method is applied to the family of CBDX models, and results and comparisons between different models, males and females, and period and cohort approaches, are presented.
