Participating longevity-linked life annuities (PLLA) are an interesting solution to manage systematic longevity risk in markets in which alternative risk management solutions are scarce and/or expensive and in which there are significant demand- and supply-side constraints that prevent individuals from annuitizing their retirement wealth. In this paper we revisit, complement and expand previous research on the design, valuation and willingness to pay for various index-type PLLA structures. Contrary to previous studies that use a single model to forecast mortality rates, we develop a novel approach based on a Bayesian Model Ensemble of generalised age-period-cohort stochastic mortality models. To determine which models received a greater or lesser weight in the final projections, we implemented a backtesting cross-validation approach. We use Taiwan (mortality, yield curve and stock market) data from 1980 to 2019 and adopt a longevity option decomposition valuation approach. The empirical results provide significant valuation and policy insights for building post-retirement income, particularly in Asian countries.