Abstract
This study aimed to investigate the effect of precarious employment (PE) on the mental health of Australians. Building on previous research, we conceptualised PE as a multidimensional construct, accounted for gender differences in the associations, and our modelling strategy addressed the possibility of reverse causality bias. Data was pooled panel data from 15 waves (2005-2019) of the HILDA survey (n = 14,237). Using PCA, we created two multidimensional measures of PE: objective and subjective. Fixed effects (FE) regression models (attending to unmeasured time-invariant confounders) estimated the change in mental health associated with a change in PE, and instrumental variable (IV) analyses (addressing endogeneity bias) obtained an unbiased estimate of effect of subjective PE on mental health (with ordinary least squares (OLS) models as baseline). For both genders, FE models showed that objective and subjective multidimensional PE both had a strong negative association with mental health (stronger for subjective PE). IV analysis indicated OLS models overestimate the relationship between subjective PE and mental health for men but underestimate it for women, providing causal evidence that subjective PE is important for women's mental health. Overall, findings suggest that targeted PE policies have the potential to provide significant population mental health gains, particularly for working women.