This paper analyzes the causal effect of citizen-initiated ballot measures, defined as initiatives and veto referendums, on state government revenue and expenditure. In the United States, several states have implemented direct democratic institutions at various times. We utilize this staggered adoption setting by leveraging a novel panel dataset spanning from 1890 to 2008. We show that the introduction of citizen-initiated ballot measures negatively affects state revenue and expenditure when using a difference-in-differences design. Significant pre-treatment effects question the causal interpretation of these findings. To overcome these identification issues, we use the novel partially pooled synthetic control method, which constructs a synthetic control group based on the pre-treatment imbalance for each treated state and across all treated units. The results indicate that introducing citizen-initiated ballot measures does not affect state finances. These findings are robust to multiple robustness tests. This research offers new insights into the fiscal implications of citizen-initiated legislation.