Essays in public economics

Anwar, Chowdhury Mohammad Sakib and Matros, Alexander and Sen Gupta, Sonali (2020) Essays in public economics. PhD thesis, Lancaster University.

Text (2020AnwarPhD)
2020AnwarPhD.pdf - Published Version
Download (1MB)

Abstract

The thesis consists of three chapters in public economics. In the first two chapters, we develop theoretical models to show how citizens’ tax evasion and the governor’s embezzlement affect public good provision. In the third chapter, I design an experiment to see how position uncertainty in a sequential public goods game affects contribution levels. In the first chapter, we consider a model that links tax evasion, corruption, and public good provision. In our model, citizens pay or evade taxes into the public fund, which a corrupt governor redistributes. Each citizen forms expectations about the amount of public goods the governor should provide. After ob- serving the actual level of public goods, a citizen punishes the governor if this level is below his expectations. We describe three types of equilibria: tax evasion, efficient public good provision, and symmetric mixed-strategy. We show that the highest expectations can lead to no free-riding (tax evasion) and the efficient level of public good provision even with the corrupt governor and without punishment for tax evasion. The second chapter complements the first chapter by relaxing the assumption of symmetric strategy. In this chapter, we consider a model with two citizens and a governor. First, the citizens decide whether to pay tax or evade. Then, Nature (an independent tax authority) audits one of the citizens randomly, and in case of non-payment, the citizen is fined and are forced to pay the tax. Third, the governor receives the taxes and decides how much public good to provide. Finally, after the governor’s decisions, citizens observe the amount of public good provided and express their opinion on whether the governor embezzled from the public fund or not. More specifically, the citizens guess the correct number of units in the public fund. We formulate this in a four-stage extensive form game. We characterise the Bayesian Nash Equilibrium of this game. Our main result shows that any strategy profile can be a Nash equilibrium for the right choice of the parameter. This suggests that, as a society, we could reach a particular set of outcomes if we set specific restrictions on the parameters (for example, we can set parameters such that there is no tax evasion). By assuming that citizens care about their guesses (i.e. opinions) we can refine our Nash pre- dictions. This gives us three different types of Nash equilibrium. First, we show that if the penalty for evading tax is too low, then both citizens have an incentive to evade tax. Then, we show that if the penalty for evading tax is high enough, and the penalty for embezzlement is low then at least one of the citizens pays tax and the governor embezzles whenever he has the opportunity to do so. Finally, we show that if the penalty for both embezzlement and tax evasion is high enough, we will have efficient public good provision, meaning all citizens pay tax and the governor uses the entire fund to provide the public good. Finally, in the third chapter, I design an experiment to see how position uncertainty in a sequential public goods game affects the level of the contributions. Theory suggests that in a one-shot game among a finite number of self-interested individuals, full cooperation is sustainable (Gallice & Monzón, 2019). This prediction is completely different from the previous theoretical pre- dictions. The experiment consists of two treatments. The first treatment is a sequential public goods game. The second treatment is the game with position uncertainty. The results show that the contributions, at the group level, in each treatment are statistically different.