A table giving mortality specific to each year of birth, thereby building in expected longevity improvement instead of freezing it.
A period table describes the mortality observed in a given year at every age, and implicitly assumes that a man of forty today will meet, thirty years from now, the mortality that seventy year old men face today. That assumption is wrong as soon as longevity improves, and the longer the commitment the more wrong it becomes. A generational table fixes the flaw by giving each birth cohort its own set of probabilities, projected through a mortality improvement model. The financial consequence is severe on the annuity side: underestimating improvement by a few tenths of a point a year is enough to underreserve a book by several percent, an error that surfaces over thirty years. It is symmetrical and favorable on the death benefit side, which is why one insurer can be exposed both ways and yet see only an apparent offset, since the two books share neither size nor duration. The problem solved is to stop treating mortality as a constant when it is a trend.
Britain's Continuous Mortality Investigation has published since 2009 a projection model revised every year, each vintage of which moves pension fund reserves by billions of pounds without a single retiree's situation changing: the 2016 version cut reserves across the British market by roughly 2%, taking in the slowdown in longevity gains observed since 2011.
table prospective, table par cohorte de naissance