2 The age-period-cohort model
Classical cohort analysis, or age-period-cohort (APC) analysis, is a method for
exploring time series of demographic data and for the comparison of life courses of
different cohorts. The time series generally consists of data classified by age and period
(e.g. calendar year). Sometimes data are grouped by age and cohort. Variations in the
age profiles are attributed to contemporary and historical factors. The contemporary
factors are usually referred to as “period effects” and are generally approximated by the
calendar year. The historical factors represent the influence of the past on current
behaviour or experience and are usually referred to as “cohort effects”. Cohort effects
occur whenever the past history of individuals exerts an influence on their current
experience or behaviour in a way that is not fully captured by the age variable. The
main contribution of APC analysis is that the impact of societal and technological
processes on demographic experience is conceptualized in its historical and
contemporary dimensions.
The cohort or generation is an important concept in the study of changes in societal
behaviour and experiences over time. Ryder recognized the cohort concept as the
dominant agent of change in society (Ryder, 1965), and it was no coincidence that his
similar article appeared in the sixties of the last century, at a time when the protest
generation was about to stir large changes in society. The interest in cohort analysis is
therefore particularly large when discontinuities occur in trends. Cohort analysis is
expected to reveal and quantify the impact in time of these discontinuities.
In traditional APC analysis, the contemporary factors are approximated by the current
period and the historical factors are represented by the year or period of birth. Current
period and period of birth are not causal factors in the analysis. They are crude
indications of the macro-setting that changes over time and in which demographic
phenomena are embedded. In traditional analysis, the demographic rates, measured for a
given age-group during a given period, are decomposed into an effect of age grouping
(age effect), an effect of contemporary factors (period effect) and lasting effect of
historical factors experienced by the group of people born during the same period; in
APC analysis, it is interpreted as a group of people who lived through comparable
historical or structural contexts (e.g. depression, war period, period of rapid
technological change). They may be referred to as “contemporaries”. Although the
impact of past common experiences remaining at the time of observation is likely to
differ for each member of the group, there is probably some effect that is still felt by all
members of the group. That effect is the cohort effect. APC analysis attempts to unravel
inter-cohort differences and intra-cohort variations.
APC analysis combines the two viewpoints traditionally distinguished by a
demographer when analysing demographic data. One approach examines changes from
year to year. Period analysis, as this approach is known, is particularly useful when
rapid changes occur, such as technological or legal changes that directly affect the
controllability of demographic processess, or a war or a revolution resulting in
transitory behavioural changes such as the postponement of births. The other approach,
cohort analysis, is better suited to the study of fundamental changes in behaviour such
as an increase in health conditions and life expectancy. A comprehensive treatment of
APC analysis in demographic and social research is given by Mason & Fienberg (1985).