Background Income and social security are fundamental determinants of mental health and wellbeing. Universal Credit (UC) was a large-scale reform to the UK’s welfare system, replacing six existing benefits and tax credits, and is currently paid to over 6 million recipients. UC aimed to simplify the claims process and encourage claimants to move into employment. Previous studies have suggested harms to mental health resulting from the new system, via mechanisms such as decreased benefit income, incurred debt and increased stress. Introduction was phased on a geographical basis from 2013 and was available in all areas by the end of 2018 to all new claimants, existing recipients whose circumstances changed and those who chose to switch to the new benefit. We aimed to identify the effect of UC on mental wellbeing on low-income households by treating the phased rollout as a natural experiment.

Methods We analysed repeated yearly cross sections of the nationally-representative Annual Population Surveys 2013–2019. The outcome of wellbeing was measured using ‘life satisfaction’, happiness, anxiety, and life rated as ‘worthwhile’. We examined respondents in low-income households (less than £12,000/year household equivalised income; n=249,078), comparing those living in areas where UC had been introduced to those in areas where it had not, using difference-in-difference analyses. In subgroup analyses, we tested for differential effects on under-25s, disabled people, single people, women, non-white ethnicities, people with caring responsibilities, single parents and full-time students (compared to complementary subgroups). We compared effects between Scotland, Wales and England, to take account of differences in the way UC is administered. All analyses were carried out using R.

Results Exposure to UC was associated with lower ‘life satisfaction’ (-0.07; -0.10 to -0.04), happiness (-0.06; -0.10 to -0.03) and ‘life worthwhile’ (-0.04; -0.08 to -0.01), and greater anxiety (0.09; 0.04 to 0.13) amongst low-income households. Effects on anxiety were larger among under-25s, disabled people, single people, women, people with caring responsibilities, single parents and full-time students. There were small, additional harms to ‘life satisfaction’ and ‘life worthwhile’ in Scotland compared to England.

Conclusion Consistent with studies of the effect of UC on mental health, we found evidence of harmful effects of UC across four measures of wellbeing. Effects on anxiety were exacerbated across multiple axes of vulnerability. This should inform recommendations for the migration of all remaining eligible households onto UC and future adaptations to the UC system to reduce adverse consequences for mental health and wellbeing.