The electorates' lack of information about the extent of public spending may cause misalignments between voters' preferences and the size of government. We devise a series of representative survey experiments in Germany that randomly provide treatment groups with information on current spending levels. Results show that such information strongly reduces support for public spending in various domains from social security to defense. Data on prior information status on school spending and teacher salaries shows that treatment effects are strongest for those who initially underestimated spending levels, indicating genuine information effects rather than pure priming effects. Information on spending requirements also reduces support for specific education reforms. Preferences on spending across education levels are also malleable to information.