De Vos, Susan
Working paper no. 2018-01
This study looks at the extended family household living of elderly women aged 60+ in Brazil in 2010 (35.9 percent) by whether they had had a live birth using 2010 Brazilian census data (36.8 vs. 29.0 percent). It applies a binomial multivariate logistic model containing socio-demographic, socio-economic, and socio-geographic characteristics to the two groups of women both nationally and by broad income group. It decomposes the gap into propensity differences (118.9%) and compositional differences ( 23.5%).
The study found that many characteristics such as income had similar relations with the probability of extended family living among women who had or did not have a live birth. While basic income had a negative relation, receiving pension income among lower income elderly women had a positive relation. This finding was a possible side effect of the non-contributory pension, as a similar finding did not exist in other income groups. The major difference between the two fertility groups was that factors associated with a special ‘need’ for informal support (such as having a disability or being very old) were more important for the co-residence of women who had never had a live birth.
For both long-standing cultural as well as politico-economic reasons, Brazilians may expect a network of kin, not just biological children, to care for and sometimes share, residence with elderly family members AND they have developed a non-contributory pension program aimed at the very poor that enables old people to control and share resources within what may be extended family households