This dissertation consists of three essays on public good provision.

The first chapter develops a model of charity’s choice of fundraising method under two dimensions of asymmetric information, quality and purpose. The main implication is a separating equilibrium where higher-quality charities choose to distinguish themselves by using a traditional fundraising method, while lower-quality ones exploit a low-stakes, take-it- or leave-it, ``checkout’’ method. An empirical application reinforced that charities of lower quality are more likely to adopt the checkout method. Despite this, consumers still choose to give in the equilibrium, due to the small requested amount of checkout donations, which disincentivizes serious thinking. Although exploited by lower-quality charities, the checkout method, along with purpose uncertainty, has the potential to alleviate the free-riding problem associated with public good provision and is, therefore, welfare improving.

The second chapter studies why corporations donate to charities and

how their donations affect social welfare. I propose that firms make donations out of an image reason. In a model where two firms compete with each other, charitable donation could attract consumers and also signal firm overall social responsibility. I show that there exists an equilibrium where the high responsibility firm overdonates,

resulting in a donation level closer to the socially optimal

one. This leads to higher consumer welfare due to higher private good

consumption as well as higher public good consumption when overdonation is prominent. Overall social welfare is enhanced. Empirical results support social image as an incentive for firms to donate.

The third chapter examines people's marginal willingness to pay for a change in local public good provision. We use a fixed effects hedonic model with MSA level data to study the effect of crime on local housing price. We explore the 1990s crime drop and use abortion data in 1970s and 1980s as an instrumental variable based on \citet*{donohue2001impact}. One result we find is that a decrease in murder of 100 cases per 10,000 people increases housing price by 70\%. We further translate this result into a value of a statistical case of homicide, which is around 0.4 million in 1999 dollars.