1997 or 1996, depending on the measure of low income, - have shown a particularly
favourable evolution of this cohort’s relative position.
The cohort of people born from 1930 to 1939 is always the second less poor, but depending
on the poverty criterion that is considered, its situation may not improve during the period.
The other two cohorts sometimes interchange their position as the poorest cohort, although
the oldest one is dominant in that position. Their situation does not improve during the
period unless the criterion of 66% of the median of 1994 is used.
When analysing differences in low income incidence by household type14, we can see that
retired people living alone are the ones that are worse off ( See Figure 4-3 and Table 4-3).
That is true both for men and for women. Nevertheless, the gap between the relative
position of that household type and the position of the next poorest household type - couple
with no children - is especially large for women. Men living in a couple with no children
used to show considerably smaller poverty incidence rates, but that changed in the last
years of our sample. For men, the first two worse off categories converged.
Men living with children (without a wife) have shown the most irregular evolution, with the
largest incidence rate taking place in 1997 (47% using two of our poverty definitions) and
the lowest incidence rates in 2000 or 2001 (5% in 2001 using the same two poverty
definitions).
For women, the other three household types show rather similar levels.
Our results are robust to the choice of the poverty criterion.
5. Income effects in the transition to retirement
The analysis in section 4 revealed a higher incidence of poverty among retired individuals.
That evidence has different implications if it results from the fact that retired and not-retired
individuals have different characteristics or if it arises from the old age social protection
system that does not cover retired individuals from the risk of poverty.
If the first case is true, waiting is the answer as the normal change of the composition of
retired and not-retired will solve the problem, but if the second is true some political action
14 We identify 5 different types of households a retired person may belong to: 1) single person household, 2)
person living with children, 3) couple with no children, 4) couple with children, and 5) others.
13