The GP framework is Bayesian, oering rich uncertainty quantication. The model produces mortality
curves smoothed over multiple dimensions, as well as credible intervals which quantify the
uncertainty of these curves. This is generated for in-sample smoothing and out-of-sample forecasts.
In their basic form, the latter forecasts are Gaussian, allowing for a simple interpretation of the
uncertainty by the actuary. Moreover, the GP model is able to generate stochastic trajectories of
future mortality experience. We demonstrate this projection over both age and calendar year, but
the GP model can be consistently applied over higher dimensional data as well. From this, full
predictive distributions for annuity values, life expectancies, and other life contingent cash-
ows can
be produced. Such analyses can provide core components of stress testing and risk management of
mortality and longevity exposures.