12. Period or cohort effect?
A full analysis of a possible cohort effect in the widespread mortality decline is yet to be made. Instead, the time series in the countries which have been experiencing sudden changes in trend are examined in the following in order to see whether the changes have taken place in different age groups simultaneously or with a time lag which could be indicative of a cohort effect.
The countries in which the sharpest breaks in trend have occurred are Japan, Finland, West Germany, Norway and Czechoslovakia. Instead of annual series which proved too unsteady, we have used quinquennial averages and calculated the percentage change from one to the next. These are shown in Table 14 for the four 5-year age groups below age 100. The development is illustrated in Figure 8 but without the group 95-99 which frequently showed rather erratic values. The following is a commentary of the findings.
M ales. A moderate decline in the 1950s came to a stop around 1960 after which began an ever more precipitous fall. Through all these changes all age groups run parallel though the oldest group only approximately.
Females. A slow initial decline stopped around 1960 after which began a sharp drop simultaneously in all age groups, accelerating till about 1975 after which it has continued apace.
Males. A slow decline in the 1950s turned into a rise and then, around 1970 into a steep decline which slowed down in the 1980s. All these stages were simultaneous at all ages.
Females. Similar development except that instead of an increase there was a stagnation in the 1960s and the drop after 1970 was even steeper than for men. All changes were closely parallel at all ages.
After a slow improvement during the first two decades, a sharp fall began simultaneously for both sexes and all age groups around 1975 and it has continued until now.
Sharp drops have alternated with relative slowdowns and even an increase around 1960, always simultaneously for both sexes and all ages though with always a more favourable development for females.
Males. Initial improvement which may be seen as return to normal after the war, was followed by an increase in mortality until the last quinquennium brought about a sudden decline at all ages.
Females. The early improvement was more pronounced and lasted longer than with men. It was followed by a period of approximate stagnation and finally a sharp drop, similarly at all ages.
To sum up, the development in any given country and for each sex has been closely simultaneous in all age groups, particularly the three younger ones which display the steadiest rates. Every important change in trend, whether up or down, has taken place virtually at the same time at all ages without any indication of time lag at older ages. It may be noted in particular that the time of onset of a rapid mortality decline has been simultaneous in the different age groups and therefore has not been produced gradually by supposedly healthier cohorts moving into higher ages. The existence of cohort factors is not precluded but it is obvious that their role, if any, has been only secondary and that the recent epoch-making decline in old age mortality has been caused by period factors with immediate effect.
Updated by V. Castanova, 1 March 1999