higher than social assistance for the rest of the population has led to the poorest decile
being better off after retirement.
To retire early is particularly and persistently connected to increases in poverty [Haveman,
Holden, Wilson and Wolfe (2003)]. Bardasi, Jenkins and Rigg (2002) find that to be true
for women.
Kingson and Arsenault (2000) stress the diversity of risks in the transition from active or
part-time working status to retirement, having found that there is room to differentiate the
income risk intensity in retirement transition by some sub-groups of the population as Afro-
americans, Hispanics, low-income earners, unmarried individuals and unhealthy early
retirees.
In our paper, we base the identification of retired individuals on the self-report of survey
respondents. This allows only the classification of someone as retired or not retired,
therefore not admitting the discussion of whether retirement is a gradual process or an
abrupt transition. That is an interesting related topic whose answer depends on the country
under consideration. In some countries -like the USA [Hungerford (2003), Grad (1990)] -
retirement is a long process, whereas in others - like in Germany [Hungerford (2003)] -
retirement is a rather definite point in time, since withdrawal from the labour force is
usually complete from the beginning, and this coincides with the receipt of retirement
income. With relation to the UK, Disney, Grundy and Johnson (1998) conclude that
retirement tends to be a one-time process, whereas Bardasi, Jenkins, and Rigg (2002) reach
the opposite conclusion, although more when considering women than when considering
men.
The study that is closest to ours is Bardasi, Jenkins, and Rigg (2002). Both studies analyse
the association between transitions into retirement and the probability of becoming poor,
considering different definitions of low income and using data from the ECHP (European
Community Household Panel). Their research is aimed at Great Britain, whereas ours is
aimed at Portugal.
The use of income data as a proxy for economic welfare has two limitations. In the first
place, when comparing income before and after retirement there is the danger of
misinterpretation if a change in needs is not taken into account. After retirement, people