Older people report much less hardship than younger people in a range of contexts, despite lower incomes. Hardship indicators are increasingly influential, so the source of this age gradient has considerable policy implications. We propose a theoretical and empirical strategy to decompose the sources of this relationship. We exploit a unique feature of the Household, Income and Labour Dynamics Australia (HILDA) survey, which collects reports of hardship from all adult household members. This facilitates within-couple estimates, allowing us to identify age-related reporting bias. The majority of the raw age-hardship gradient is explained by observed resources, particularly wealth and home ownership. One third of the relationship is explained by unobserved differences between households, which we interpret as age-related behavioral choices. Reporting error does not appear to contribute to the age gradient.