mortality projection

[Harleen Anand]

There is a line in the study material which is as follows:
it is possible to work with the age and cohort but the cohort approach makes heavy data demands and there is a largely insolvable problem that the recent cohort have their history truncated by the present day.
firstly what is this unsoluble problem and what does it means by their " recent history truncated by the present day "

Thanks

 

[Andrew Martin]

Hello

When analysing mortality by age, we should have lots of data available (lives aged x at some point in time) to help estimate age effects. When analysing by period, we should also have lots of data available (lives alive at particular points in time).

However, when analysing by cohort, we're considering more specific groups of lives, say those born in a particular year. For example, say we're interested in the cohort born in 2015 and the specific mortality related to these lives. Well, we only have 5 years of past data for these lives so it is difficult to understand their specific cohort effect (how this cohort may differ to others) with such a small amount of data.

There's also nothing we can really do about it, we just have to wait for the lives to get older so that we have more data on their particular cohort.

Hope this helps.

Andy