Background There has been little investigation of changes in socioeconomic measures and mental health (MH)/illness over time within individuals using methods that control for time-invariant unobserved confounders. We investigate whether changes in multiple socioeconomic measures are associated with self-reported MH using fixed effects methods to control for unobserved time-invariant confounding.

Methods Data from three waves of a panel study with information on MH, psychological distress, labour force status, household income, area and individual deprivation. Fixed effects regression modelling was used to explore whether changes in socioeconomic exposures were associated with changes in MH. We also compared increases and decreases in exposure with changes in MH using first difference models.

Results Respondents who moved into inactive labour force status experienced a 1.34 unit (95% CI −1.85 to −0.82) decline in SF-36 MH score and a 0.50 unit (95% CI 0.34 to 0.67) increase in psychological distress score. An increase in individual deprivation was associated with a 1.47 unit (95% CI −1.67 to −1.28) decline in MH score and a 0.57 unit (95% CI 0.51 to 0.63) increase in psychological distress. Increasing and decreasing levels of individual deprivation were associated with significant changes in both outcomes.

Conclusions This paper suggests that moving from employment to inactivity and changes in levels of individual deprivation may be more important for short-term MH outcomes than changes in household income or area deprivation. Providing short-term social and economic support for those experiencing financial/material hardship should be considered in interventions to reduce inequalities in MH.