Order, order!

(Apr 6, 2011)

Mortality improvements can be analysed in a number of ways.  A common desire is to want to separate mortality improvements into components for period and cohort.  However, this is much trickier than it seems, as we shall show here.  In particular, the order in which calculations are performed can be very important.

Figure 1 shows the mortality improvements derived from a two-dimensional smoothed model, identical to the ones presented by Richards, Kirkby and Currie (2006).  The vast majority - 97.3% - of the improvements are positive, and there is a pronounced diagonal pattern reflecting the tendency for mortality improvements to track year of birth, or cohort.  Indeed, the diagonal cohort patterns are…

Read more

Tags: mortality improvements, period effects, cohort effect

Find by key-word


Find by date


Find by tag (show all )