Estimating the effects of household income on weight in Australian adults

Abstract

This paper employs data from the Household, Labour and Income Dynamics in Australia surveys to estimate the effects of gross household income on body mass index (BMI), with a focus on adults from low- to medium-income households. The potential endogeneity of household income is controlled for using two separate methods. Fixed effect panel estimations are conducted with individual data from 2006 to 2010, and instrumental variables estimations are performed on individuals from low-income households in 2009, using Australian government bonus payments from the 2008-2009 financial year as an instrument for household income. Fixed effect results suggest that there is no significant relationship between household income and individual BMI after controlling for other variables. In examining individuals from low-income households, although fixed effects estimations indicated small positive relationships, instrumental variables results suggest that household income has no causal effect on BMI. This implies that positive relationships found in the fixed effects estimates may be due to endogeneity caused by time-varying omitted variables or reverse causality. These results were robust to analysis by gender and weight classification, corrections for self-reporting error and examining change in BMI rather than BMI level.