This paper notes that while most literature on intergenerational transfers of wealth uses a nuclear family (parents and children) model, that model is not typical of much of the developing world, where the extended family is the prevalent household structure. The author, drawing on her extensive research in Philippine villages in the mid-1980s, looks at how resource allocations and wealth transfers (principally as they relate to land and investment in education) are carried out in three-generation extended families and how grandparents, especially grandparents who live with or in near proximity to the household, influence decisions parents make regarding transfers of education and land to children. Overall, grandparent-influenced transfers of land favor grandsons, but nonland wealth transfers result in greater investment in education for granddaughters.

The Study
In 1989 the author did a retrospective survey of 344 families in five villages (two in Central Luzon and three in Panay) originally surveyed for the International Rice Research Institute (IRRI) in 1985. The 1989 survey sought information on three generations—grandparents, parents and siblings, and (grand)children. Parents (the respondents) were asked about the education and landownership of their parents and in-laws, their own education and inheritances, and the schooling of and proposed bequests to their children. The survey matched 170 sets of grandparents and parents with 795 children over 18. The average birth year for grandparents was 1909; for parents, 1939; and for children, 1959. While grandparents typically did not live in the same house with their adult children and grandchildren, except during their final years, 30 to 40 percent lived and died in the same village as their adult residence, allowing the author essentially to equate proximity with coresidence to establish her model of the three-generation extended family she analyzes. The educational attainment portion of the study used the whole sample. The analysis of land bequests covered only a subset of 68 families that had decided on specific size bestowals to 353 children above 18.

The author uses ordinary least squares (OLS) estimates of grandparent, parent, and child characteristics—gender, age, wealth (parents and grandparents), and levels of education (parents and grandparents)—in regression analysis to establish how the extended family's decisions about passage of wealth—essentially land—and investment in education will be made.

In the Philippines, overall, grandparent-influenced transfers of land favor grandsons, but non-land wealth transfers result in greater investment in education for granddaughters.

Findings
The paper finds that in the rural setting where grandparents are coresident or quasi-coresident, grandparent wealth affects the allocation of land to children. To an even greater degree, grandparent proximity affects the level of education attained by children. Typically, male children inherit almost twice as much land; female children get more education, an average of 9.54 years for daughters compared to 8.53 years for sons. The author suggests that this indicates Filipino parents and grandparents consider both equity and efficiency goals. Agriculture is masculine work while daughters, who are socialized to be loyal to parents and use their education to get jobs outside the village, can be counted on to remit earnings home. This land-education allocation pattern is consistent with a risk- diversification strategy for parents.

The parents' role in these decisions appears to be greater, but grandparent participation is not small. Even the fact that coresident or quasi-coresident grandparents can relieve the mother of child-care responsibilities and allow her to enter the labor market or allow older girls to continue schooling rather than having to care for younger siblings influences allocation decisions.

While the author's findings are supported by the mathematical analysis of factors she employed, she acknowledges that more study is needed of the exact mechanism by which grandparent coresidence affects investment in children.

Policy Implications
Much research supporting rural development policy in developing countries looks at the family resource allocation process. Most often this research employs a nuclear family model to examine the husband-wife bargaining process that determines this allocation and how it affects children. However, as this study argues, the nuclear family is not typical of rural areas in many countries. Policymakers need to take account of this fact both in the framing of research and in the design and application of rural development policy.