Answers are grounded in this paper and cite their sources.
High-dimensional mortality models are accurate but complex, while parsimonious extrapolative models can behave poorly at extreme ages. We propose the coupled-mode model (CMM), a two-state framework for old-age mortality forecasting. Under a Gompertz hazard, modal age at death, M, and remaining life expectancy at the mode, e(M), satisfy a modal anchor identity linking e(M) to the inverse ageing rate. M and lne(M) evolve jointly in a cointegrated vector error correction model. Across 50 population–sex combinations from the Human Mortality Database, Johansen-family tests support non-zero rank in 39 cases. In 2011–19 validation, CMM’s pooled e65 mean absolute percentage error is 3.28 per cent, vs 1.24–2.11 per cent for five benchmarks; CMM is best in three populations. Interval coverage is adequate at short horizons but deteriorates over decades. CMM provides a parsimonious, biodemographically interpretable decomposition of senescent mortality dynamics.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: