Background Political decisions, including the generosity of welfare resources, can have implications for population health and health equity, especially in times of economic crises. Greater welfare generosity is hypothesised to relate to better mental health, especially among more disadvantaged groups. Whether socioeconomic inequalities in mental health have widened in response to declining welfare spending is not well known. This study first explores whether overall mental health and educational inequalities in mental health changed following the economic crisis. Second, it examines social protection expenditure as a mechanism for improving population mental health and reducing educational inequalities in mental health.

Methods Individual-level data (N = 49,211, aged 25 to 64 years) were taken from the 2006 and 2012 waves of the European Social Survey, a cross-sectional survey which uses random probability methods. Country-level (N = 20) social protection spending (including unemployment, disability, families, housing and social exclusion) measured in Purchasing Power Standards per inhabitant were extracted from Eurostat and adjusted for need using the inverse of the employment rate. Mental health was assessed using the Centre of Epidemiological Studies Depression Scale (CES-D 8). Data were analysed using random-slope multilevel linear models in Stata MP/12.1. Potential confounders were controlled for (such as age, immigrant status and Gross Domestic Product) and models were stratified by gender. Slope indices of inequality (SIIs) were calculated and the interaction with survey year was tested. Cross-level interactions between social protection spending and education were explored.

Results Mean depressive symptoms declined in 2012 among men (b = –0.21, 95% CI:–0.31 to –0.12) and women (b =  –0.34, 95% CI: –0.44 to –0.24). Higher education was associated with fewer depressive symptoms among both men (SII = –1.71, 95% CI: –2.20 to –1.23) and women (SII = –2.50, 95% CI: –3.04 to –1.97). Among men, educational inequalities in depressive symptoms widened in 2012 (p = 0.048 for interaction with year), with no change observed among women. Sensitivity analyses restricting the sample to those aged 25 to 59 years revealed this result was confined to men aged 25 to 64 years. In both 2006 and 2012, expenditure on social protection was unrelated to mental health. Greater spending on social protection was associated with narrower educational inequalities in depressive symptoms in both years and among both genders.

Conclusion Increased welfare generosity was related to narrower educational inequalities in depressive symptoms. Austerity measures have the potential to widen inequalities and exacerbate  depression among more disadvantaged groups. The main limitation of the study was its cross-sectional design.