(*) Chargé de recherche à l'INED
(**) Allocataire de recherche à l'INED et doctorante à l'Institut d'Etudes Politiques de Paris
In a country such as Indonesia, where public transfers are limited, children often support their elderly parents, possibly deciding to live with them. Martin (1989) reports that about 75 percent of the elderly in Asia live with at least one of their offsprings.
The determinants of coresidence have been studied by Da Vanzo and Chan (1993) and Slusher (1996), both using data from Malaysia. They show that opportunity costs are a major determining factor for living arrangements. In particular, Slusher shows that coresidence with parents is less likely for children with higher wages. However, Slusher's results were derived from the analysis of independent parent-child pairs (that is with information on only one child per family), which neither allows to control for family fixed effects nor to analyze living arrangements as resulting from a collective familial decision involving all children.
It is precisely this familial approach that we favour in this paper in which we try to answer to the following question : do choices of living arrangements are part of a rational and economically efficient strategy of the family ?
The theoretical model that is developed in this paper combines inter-household transfers and living arrangement patterns and focuses on how is chosen the child who will cohabit with the parents. In practical terms, for a set of siblings with elderly parents, several living arrangements are possible (parents may live alone or live with the child $A$ or the child $B$, etc.) but most of them are not Pareto optimal because total family resources are not maximized. For example, let's assume that parents are better off if they live with a child, that both children $A$ and $B$ care for their parents, but that they would suffer some opportunity costs if deciding to live with their parents. If there is no cooperation between $A$ and $B$, they might both decide not to live with their parents while a situation with $A$ living with the parents and $B$ sending money to $A$, as a compensation, would make everybody better off. In such a case the outcome would obviously not be Pareto optimal at the family level which would suggest that children are not committed in a cooperative relationship.
In the paper, we use the first Indonesian Family Life Survey from 1993 which provides an unusually rich information on intra-familial transfers in order to investigate this ''cooperativeness'' of the inter-sibling relationship. Since the data from the Indonesian survey reports information on several children, we are able to control for fixed effects of each family. The econometric estimations, based on Chamberlain's method (1980), show that parents are more likely to live with their less educated child and that the other children are more likely to help their sibling cohabiting with the parents. In other words, siblings behavior is consistent with what would follow from a cooperative agreement. The care of the parents is achieved by the less educated child, in order to minimize the opportunity costs, while the other children send money to this child as a form of compensation.
While these results are preliminary, they do shed light on an important theoretical question regarding the motivation for transfers. They suggest that siblings should not be seen as independent or competing individuals but rather as a cooperating group. In fact, the results shown here challenge some classical theoretical models on intergenerational transfers, such as the ``Strategic bequest motive'' model of Bernheim, Schleifer and Summers (1985).