This paper tests, using data from South Africa and Pakistan, two major implications of the unitary household model, namely, that (a) each individual pools the various components of her/his non labour earnings, and (b) men and women pool their non labour earnings between themselves. The study uses a three stage least squares procedure that, besides recognising the endogeneity of all the income variables, allows for simultaneity between all the income and expenditure equations. The study finds that men and women are much less likely to pool their transfer receipts than other types of income. This paper, also, investigates the crowding out of private transfers by public transfers in both countries. While it finds no crowding out in Pakistan, it reports strong evidence of such crowding out in South Africa in 1993/94 and that too for the poor but not the non poor households. However, the negative impact of social pensions on private transfers in South Africa seems to have weakened over 1993-98. This study also finds that, in 1993/94 though not in 1998, social pensions had a negative impact on earned income of members, especially of females, in the household. The social pensions scheme in South Africa, as prevailing in 1993/94, was not as generous towards black households or as redistributive towards the poor, as commonly believed. However, the pensions scheme seems to have changed in both these respects over\n1993-98.