Electoral accountability for corruption has been shown to be generally weak. However,
we still know little about why this is the case. I propose a novel argument to explain this
puzzle. Just how weak accountability for corruption is depends on the salience of
corruption to voters, and how salient corruption is depends, in turn, on an individual's
income. I test this explicitly political economy argument in conjoint experiments and
with large-N observational survey data to measure the salience of corruption and its
effect on vote choice, moderated by individual income. Results from India and the
United Kingdom imply that there is a purely material self-interest side to corruption
voting. Voters do not punish politicians for corruption when corruption is not salient to
them, and this depends on whether they are rich or poor.