Sullivan's approach is similar to Wolfbein's approach [] with a ``Working life table''. Discussions of the same type as current discussions on ``active life expectancy'' raised in demographic litterature [,,,], [,,,], until Jan Hoem [], described a correct computation of Working life tables from longitudinal survey in Denmark (1972). Demographic implications in particular for projecting female activity rates for France, have been presented in [].
Sullivan's approach consists in computing cross-sectional prevalence of dependancy by age from a cross-sectional survey (one round survey) and to put them into a pure period life table (obtained from another data source, such as vital statistics and censuses).
Using a cross-sectional survey which involves the different past histories for different generations, in order to compute a period indicator, has the major drawback of affecting the validity of the resulting indicator. This is as if we were to compute life expectancy from the proportions of survivors in each generation as shown on figure 2 for Japan in 1982,
[p]