The third addressed question regards the choice of a variable of resources or income. We
choose to consider as variable of resources the income per equivalent adult, the household
remaining the unit of measure, as previously said. We have adopted the OECD modified
equivalence scale to allow for comparability of incomes between households with different
dimensions and compositions. The household income is expressed in real terms
(base=1994) and we have also chosen to consider the current income per equivalent adult
(that is, the available income at the month prior to the interview) instead of annual net
income (which refers to the year prior the interview), as this lag between the reference date
for income and, namely the dating of retirement, might introduce some noise in the
analysis. Current income per equivalent adult is calculated from the household current net
income after deduction of direct income taxes and contributions for the social security
system. Household income is defined as the sum of cash income from all sources: labour
market incomes from employment and self-employment, private investments and savings
income, public occupational and private pensions and other net cash benefits from social
security system11.
As much attention is put on the analysis of poverty incidence, four monetary poverty lines
are used throughout the paper: three contemporary poverty thresholds and an "anchored"
poverty line. They are the following: a) the first quintil of the income per equivalent adult
distribution, b) the 33rd percentile of the income per equivalent adult distribution, c) 60% of
the median of the income per equivalent adult distribution and d) the "anchored" threshold,
as 66% of the median of the income per equivalent adult distribution on wave 1, 1994.
Throughout the paper we use different subsets of the ECHP panel. In section 4 we use the
whole sample to analyse the incidence of poverty among retired people in comparison with
other groups of the population and to analyse the incidence of poverty in different groups of
retired.
In section 5 we concentrate on those that have retired during our sample period. The
selected sub-sample considered in the dynamic analysis has 974 individuals (481 men and
493 women) aged 50-69 years at the time of entering the panel database. These individuals
have experienced a transition process to retirement state within the time window of the
11 This income decomposition is only available on the annual net income variable and not on current income
data, the income variable which we will use in this paper.
11