OBJECTIVE: To explore and describe quantitatively the effect over time of unemployment on food purchase behaviour and diet composition.

DESIGN: Longitudinal data from 2008-2012, with monthly food purchase data aligned with register data on unemployment measured as a dichotomous indicator as well as a trend accounting for the duration.

SETTING: A household panel which registers daily food purchases combined with detailed nutritional information and registration of the duration of unemployment at individual level. The structure of the data set facilitates the detection of effects or associations between duration of unemployment and diet composition, purchase behaviour in terms of food expenditure, and share of food purchased on offer and in discounters while controlling for important confounding factors.

SUBJECTS: Danish households of working age (n 3440) adjusted to household equivalents. We use fixed-effects econometric methods to control for unobserved heterogeneity.

RESULTS: In the short run, unemployment led to substitution in favour of discount stores and increases in food expenditure and in consumption of saturated fat, total fat and protein due to increased consumption of animal-based foods. In the medium run food expenditure declined together with consumption of fresh animal-based foods and saturated fat, total fat and protein. In the even longer run these nutrients were substituted by carbohydrates and added sugar.

CONCLUSIONS: Unemployment has a substantial influence on diet composition, but effects vary with duration of the unemployment period, which may have potential health implications. This ought to be taken into consideration in evaluations of existing reforms and in future reforms of welfare systems.