Fiscal contract theories of the state hypothesize that government performance affects tax collection and that institutions that foster representation and accountability link taxes and services. These propositions have yet to be tested with a causal research design and disaggregated data. In this paper, we use a quasi-experimental research design with randomized audits from Brazil to assess whether revealed corruption and other metrics of government performance affect municipal property tax collection. We find that revealed corruption robustly reduces property tax revenue; revealed performance on many other dimensions does not. We also find that while the structure of fiscal institutions (participatory budgeting versus budgeting via elected representatives) has no effect on the amount of corruption revealed or the quality of performance, revealed corruption increases the probability that a municipality will adopt participatory budgeting in the future.