The structure of intra-household allocation is crucial to knowwhether a transfer from a rich household to a poor one translates into atransfer from a rich individual to a poor one. If rich households are moreunequal than poor ones, then a progressive transfer among households reducesintra-household inequality, hence inequality among individuals. Morespecifically, if the part of the couple's expenditures devoted to goodsjointly consumed decreases at the margin with the couple's income as well asthe part of private expenditure devoted to the disadvantaged individual,then the Generalized Lorenz test is preserved when passing from thehousehold to the individual level. This double concavity condition isnon-parametrically tested on French data. Using three definitions of publicexpenditures and two ethical rules, the data do not reject the doubleconcavity condition and support the thesis that for purposes of welfarecomparisons across individuals, the structure of intra-household allocationscan safely be ignored.