Retrospective voting is one of the most important mechanisms for democratic control. However, it hinges on voters’ ability to attribute economic changes to government decisions. Existing scholarship disagrees as to whether retrospection is ‘blind’, but has focused mostly on voters’ responses to politically ‘irrelevant events’ like tornadoes and shark attacks. Such events, I argue, may not say much about voters’ ability to attribute responsibility for the bulk of their everyday economic experience. To examine this question, I use a microsimulation model on survey panel data to decompose voters’ pocketbook changes into a policy-induced and a non-political component. By linking this to various survey measures, I track how these pocketbook components relate to economic evaluations and vote intention over time.