panel, that is, the subset of individuals belonging to that age group who, at the time of
entering the panel, were classified as “not-retired” and during the panel time are observed
to make a transition into retirement.
In section 6 we further restrict the analysis considering, from those that have experienced
the transition to retirement during the sample period, only those that were not poor in the
year prior to retirement and that are either poor or not poor after retirement.
4. Cross-section comparisons
Regardless of the definition of poverty, its incidence is always larger for retired than for
not-retired persons (see Figure 4-1 and Table 4-1). The group with the lowest poverty
incidence rate is that of workers.12 The first half of the period of the analysis has been
especially advantageous to retired men, since they were the group whose situation
improved the most.
Retired and not-retired women usually show a larger proportion of individuals in poverty.
That changes when only workers are considered. Based on the cross-section evidence,
feminine workers show slightly smaller poverty incidence rates than masculine workers.
The different results obtained when analysing workers and not-retired individuals must be
due to the existence of a more significant part of women than of men who have no paid
activity, and are not retired.
When using the median criteria, the evolution of income has been favourable to men,
between 1994 and 2004. This means that equivalent income has evolved in a way that men
have become better off.
For retired workers, older cohorts tend to be more subject to poverty than younger ones13
(see Figure 4-2 and Table 4-2). The conclusions we may reach from the cohort analysis do
not change much if we consider annual or current income.
The cohort of people born since 1940 is always the one with the smallest incidence of
poverty, regardless of the poverty line that we consider. The first years in the sample - until
12 “Workers” is a subset of “not-retired”. “Not-retired” includes unemployed and some inactive population.
13 In our study we use four different cohorts: the first one with individuals born since 1940, the second cohort
with individuals born from 1930 to 1939, the third cohort with individuals born from 1920 to 1929, and the
oldest cohort with individuals born until 1919. The age composition of these cohorts is naturally different.
12