The methodology by Lazear and Michael (1988) is used to decompose household expenditures into that for adults and children. Some specific estimation procedures are modified and cross section-time series (panel) data are used to control for household level heterogeneity. In addition, a new and approximate test for the estimated ratios is applied. The empirical results indicate that even though per child and total child expenditures are increasing with income the relative expenditure on children falls with it. The ratio of child to adult expenditures for female-headed households is less than for male-headed households. While households with a larger number of children expend more on children in per capita terms, the number of adults is negatively related to expenditures per child. Siblings of the spouses have a significant impact on the expenditures on children; particularly, siblings of the wife seem to compete with her children. The completion of primary education by both spouses positively affects relative expenditures on children. Intergenerational effects, through education and wealth, are also important. Pre-marriage wealth, particularly for the female spouse, positively affects allocations to children. Length of marriage and the existence of a written marriage contract increase relative expenditures on children. Both measures reflect stability of marriage indicating that ‘optimal’ matching in the marriage market is an important determinant of intra-household allocations. Individual level fixed effects regressions indicate that weight-for-height z-scores of children are more correlated to the estimated expenditures on children than with total household expenditure. This result on the one hand underscores the importance of intra-household allocations, and on the other shows that the estimated expenditures on children and the underlying assumption used to derive them are not off the mark.