More income, less depression? Revisiting the nonlinear and heterogeneous relationship between income and mental health.

This paper uses a large-scale nationally representative dataset to examine the nonlinear effect of income on mental health. To investigate their causal relationship, the exogenous impact of automation on income is utilized as the instrument variable (IV). In addition, to explore their nonlinear relationship, both income and its quadratic term are included in regressions. It is found that the impact of income on mental health is U-shaped rather than linear. The turning point (7.698) of this nonlinear relation is near the midpoint of the income interval ([0, 16.113]). This suggests that depression declines as income increases at the lower-income level. However, beyond middle income, further increases in income take pronounced mental health costs, leading to a positive relationship between the two factors. We further exclude the possibility of more complex nonlinear relationships by testing higher order terms of income. In addition, robustness checks, using other instrument variables and mental health indicators, different IV models and placebo analysis, all support above conclusions. Heterogeneity analysis demonstrates that males, older workers, ethnic minorities and those with lower health and socioeconomic status experience higher levels of depression. Highly educated and urban residents suffer from greater mental disorders after the turning point. Religious believers and Communist Party of China members are mentally healthier at lower income levels, meaning that religious and political beliefs moderate the relationship between income and mental health.