This thesis explores the relationship between household size and children's cognitive development in Australia's Indigenous communities. Past studies show that the number of children in a household has a detrimental effect on children's cognitive development (Becker and Tomes, 1994). Nevertheless, more recent empirical studies suggest that after controlling for endogeneity, this effect is non-negative and negligible (Black et al., 2005). To investigate the importance of household size on early cognitive development which is essential to life-long socioeconomic wellbeing, the Footprints in Time: Longitudinal Study of Indigenous Children, henceforth referred to as LSIC is used.
LSIC provides one of the first in-depth surveys of Aboriginal and Torres Strait Islander families (FaHCSIA, 2009). The data is used to estimate the effect of house- hold size on children's cognitive development. An instrumental variable (IV) approach is employed, which uses the widely observed phenomenon of parental preference for mixed sex siblings as an instrument for fertility. The IV approach, which has been proposed by Angrist and Evans (1998) is based on a pooled cross-section of the data. The second treatment of the data is as a panel for children across the five years the LSIC study was undertaken. Fixed effects (FE) and Mundlak regressions (Mundlak, 1978) are used to determine whether the relationship between household size and children's cognitive development holds.
The original hypothesis of a detrimental relationship between household size and children's cognitive development holds in the pooled regressions. However, the panel regressions diverge, suggesting there is little in the way of a relationship between household size and children's cognitive development. Thus, on the basis of the pooled IV results, targeting educational resources towards households with young children may assist in counteracting the effects of household resource dilution as resources are spread more thinly among a greater number of children.