People often vote with their pocketbooks. Is this an effective tool for holding governments accountable? A key unresolved question is whether voters reliably respond to the policy-induced component of their income changes or simply react to all income changes regardless of cause. Existing research cannot definitively answer this question due to methodological limitations: studies of single policies cannot distinguish voters responding to the income shock itself or to income changes regardless of their origin, while research on pocketbook attribution relies on potentially biased self-reports. To answer the question, I introduce a novel approach linking a large survey panel to policy microsimulation models that track how tax-and-transfer policies directly affect disposable incomes. This allows me to decompose respondents’ total disposable income changes into policy-induced and residual components, providing a direct test of what income variation drives pocketbook voting. Applying this approach to the UK in the 2010s, a case of significant policy-driven income variation, I find that voters do not hold incumbents more accountable for policy-induced than residual income changes on average. Instead, they respond to total income changes, which prove to be a poor proxy for policy-induced income changes. These findings suggest that voters fail to reliably reward and punish incumbents for policies that affect their disposable incomes. Pocketbook voting may therefore be a weaker accountability mechanism than commonly assumed.