Demographic projections: basic mechanisms and a retrospect on the French experience

Demographic projections are generally considered as more robust than all other kinds of socio-economic forecasts. But doubts are sometimes cast on their relevance, especially when strong revisions of previous projections are proposed, as done by INSEE in 2006. The purpose of this paper is to come back on the scope and limits of these projections. We first recall their methodology and a few elementary results concerning population dynamics, with a particular emphasis on how fertility and migration codetermine the long run demographic growth rate. We then use the retrospective experience of projections conducted at INSEE since 1964 to identify where are the major weaknesses of these projections but also their main strengths. This retrospective examination shows strong revisions concerning trends for total population or for the population in working ages, but without any questioning of the global ageing trend. We explain this result by isolating a concept of trend ageing due to the pure role of life expectancy. Changes in fertility levels or in migration flows lead to deviations from this trend that, in the French case, remain second order and more or less temporary. We conclude by outlining that the uncertainty that affects projections does not make them irrelevant for decision making. Decision under uncertainty is the common lot for most of our problems of collective or individual choice. The right answer to this uncertainty is to set in appropriate adjustment rules to deal with revisions that unavoidably stem from this uncertainty.