Experimental Investigations in Public Economics

There has been an explosion in the use of experimental methods since the 1980s. This dissertation applies those experimental methods to various topics in public economics like tax evasion, crowding-out, labor supply, and cooperation. In chapter one I manipulate tax rates and whether students are capable of tax evasion in the laboratory. In joint provision settings, where public goods are provided through donations and tax revenues, these manipulations provide clean inference about the effects of tax evasion on crowding-out. Moreover, I collect measures of warm glow to estimate its relationship with contribution behavior across evasion settings. I find the usual crowding-out result when there is no tax evasion, however, I find complete crowding-out when tax evasion is possible. The collected measures of warm glow do not explain these results. The second chapter is co-authored with Robert White. In that chapter students complete real effort tasks in the laboratory for a piece-rate payment and we manipulate the shapes of these piece-rate earnings schedules. These shapes are intended to capture different features of the U.S. tax code when the loss of benefits (e.g. SNAP, WIC, Section 8 Housing, and Medicaid) are accounted for. We analyze how the shapes of these piece-rate schedules (as well as other behavioral factors (e.g. loss aversion, overconfidence, etc.) affect labor supply. We find that the piece-rate schedules that imitate ``notches" (or discontinuous jumps in the budget set) do reduce labor supply, however, this behavioral response to taxation is learned across rounds. Chapter three provides the first direct study of the ``restart effect" in public goods games. The restart effect is said to occur when there is an upward pulse in contributions to the public good following a stoppage in game play. I manipulate the language of the experimental instructions as well as the length of time prior to the stoppage in game play to search for facts about what supports the restart effect. I find that language in the instructions about when the stoppages occur promotes the restart effect. Finally, Chapter four presents evidence from a non-experimental project where data problems prevented useful findings about the relationship between competition and public service quality.