Exploring household heterogeneities of the Deaton-Paxson puzzle: Evidence for Argentina

Theory predicts that economies of scale associated with the consumption of shared household public goods make larger families better off, given the same level of per capita expenditure or income. Public goods are relatively cheaper, while per capita expenditure on the private good will increase, as long as it is not easily substitutable, as in the case of food. However, Deaton and Paxson (1998) found exactly the opposite: food share declines with the number of heads, keeping household per capita expenditure constant. This paper aims to better understand the heterogeneities underlying the Deaton-Paxson paradox in food consumption, using data from the Argentinean Household Expenditure Survey (ENGH, Spanish acronym) for the period 2017/2018. We first differentiate the impact of an additional adult from an additional child on food demand, in families of different sizes. Second, we evaluate the relationship between food demand and household size on the distribution of income. Third, we explore potential associations beyond the conditional mean of food consumption. Because standard analysis focuses on average effects of family size on food demand, the existence of the paradox at the lower and upper end of the conditional food distribution remains unknown. Our evidence supports the findings of Deaton and Paxson (1998), and reveals important differences driving this food puzzle. Our results shed light on the crucial role of economies of scale in poor households.