This thesis first establishes a theoretical platform by reviewing and synthesizing the debates over the rectangularization of the survival curve and its links to the limits to life expectancy. An improved approach to normal life durations has been conceptually developed and empirically tested in relation to the limits to human longevity. Three dimensions have been proposed to disentangle the complexity of the transformations of the survival curve (1) horizontalization (2) verticalization and (3) longevity extension. The second purpose is to provide an analytical assessment of different indicators for measuring the transformations of the survival curve and the compression of mortality. The most relevant indicators, proposed and selected, are the horizontal “N” index, C50, the prolate ind...[ Read more ]

This thesis first establishes a theoretical platform by reviewing and synthesizing the debates over the rectangularization of the survival curve and its links to the limits to life expectancy. An improved approach to normal life durations has been conceptually developed and empirically tested in relation to the limits to human longevity. Three dimensions have been proposed to disentangle the complexity of the transformations of the survival curve (1) horizontalization (2) verticalization and (3) longevity extension. The second purpose is to provide an analytical assessment of different indicators for measuring the transformations of the survival curve and the compression of mortality. The most relevant indicators, proposed and selected, are the horizontal “N” index, C50, the prolate index K and the prolate “N” index, four standard deviations above the mode for measuring the maximum life span within normal life durations and the mode. Building on the work of Comfort (1964) “scalar expansion of the survival curve” and the last work of Kannisto (2001) “invisible wall”, testable hypotheses are formulated and tested for examining the limits to normal longevity. A complete set of “Normal” indicators (i.e., late mode (M), standard deviation above the late mode (SD(M+)), four standard deviations from the late mode (4SD(M+)) indicating minimum life span and maximum life span in the normal distribution) has been proposed. Based on the normal approach, I also introduce the impact of the elimination of cancer and heart diseases on normal life durations rather than just examining its impact on the mean longevity. More importantly, this approach provides an alternative to estimate the lower bound of deaths related to the ageing process.

Using newly constructed life tables for Hong Kong 1976-2001, the empirical results reveal that the survival curve is more horizontal than vertical and the increase in the maximum normal longevity becomes stagnate, particularly for females in the period of 1996-2001 remaining at age 117.7. The findings are in favour of hypothesis 2 - a distant invisible wall in which M has been increasing monotonically, whereas SD(M+) and 4SD(M+) decrease at a steady pace over time, denoting that more Hong Kong people die at similar advanced ages under normal life durations unaffected by premature deaths. The combination of the increase in M in synergy with the decrease in SD(M+) accentuates the increase in the age corresponding to the minimum life span, a sign of postponed ageing process. Symmetrically, the fall in SD(M+) exactly compensates the rise in M resulting in a constant or slowly mounting maximum life span, disclosing an invisible wall. Following this pattern, the survival curve will be completely rectangular when a convergence takes place in which the maximum normal longevity reaches 121 years, M and the minimum life span converge to the maximum normal longevity and SD(M+) becomes zero for both sexes in Hong Kong by the year of 2126.

Key words: Rectangularization, Survival curve, Indicators, Limits, Human longevity, Normal life durations, Hong Kong