Q1: To play devil’s advocate, do mortality improvements stem from a large number of small events, rather than a small number of large events?
We should certainly not preclude this but note that all catalysts are not equal in magnitude. If you accept that as a non-controversial statement, it then follows that some will have more impact than others - and it is the major ones we are principally seeking to identify.
The "smoking story" is littered with a number of catalysts of varying magnitude, as shown here.
Furthermore, history points to a combination of both (see other Past Longevity Catalysts) and to the extent we are able to point to / identify possible future catalysts;
(a) this certainly makes us no worse off and possibly materially better off in adding to our toolkit as managers of longevity risk.
(b) this may be symptomatic of a change in this mix towards a greater proportion of large catalyst events having a significant future impact
The inherent clear difficulty in being able to capture all such possible events should not mean we choose not to pursue the clearly identifiable ones. Indeed, the imperfect nature of being aware of only some catalysts serves to illustrate the even greater imperfections of the current status quo in which they are all largely ignored.
Q2: What about the parallel concept of Mortality Cataysts?
There are clear parallels and this is something the working party may look to develop in future. This would have the merits of
addressing the potential issue of assumption bias whereby drivers underlying increased mortality improvements are given greater consideration than triggers that cause the rate of improvement to stabilise or decrease.
a simple corresponding definition: A Future Mortality Catalyst is defined as an event with the following properties
(i) An event which has not yet occurred
(ii) Its occurrence or non-occurrence should be unequivocal and take place at a unique point in time.
(iii) Its occurrence is widely expected to lead to a demonstrably significant and negative effect on the overall life expectancy of the associated population.
A possible example is "the abolition of the NHS".
There are some key differences however vs Longevity Catalysts. In particular, mortality risk is very different in nature; the time lag between catalyst incidence and revalation within the data is typically much shorter or even zero (for example, the onset of a pandemic).