Essays in dynamic public finance

Abstract

An upcoming trend in public finance is applying economic theory to understand how private agents' incentives, expectations, and perceptions affect the efficiency and sustainability of government policies. This doctoral dissertation presents novel studies that fall into this category. The first three chapters examine many aspects of tax amnesties from this perspective, while the last chapter studies intergenerational risk-sharing under limited enforcement. The first chapter introduces a selection of stylized facts about tax amnesties by examining the recent state-level tax amnesty experiences in the US. We provide observational evidence on the heterogeneity of tax amnesty implementations among US states. The heterogeneity is also persistent throughout the last four decades. We show that a few states declaring amnesties very rarely in the 80s and 90s start to implement amnesties much more frequently in recent decades. We also observe that the tax amnesty declarations of US states cluster around the US recession periods. The second chapter introduces a theoretical framework to investigate the strategic interaction between a government and taxpayers in an economy. Our model predicts four factors that make a tax amnesty more likely in an economy: high personal income; high tax rates; low political cost for declaring an amnesty; and low audit rates. We examine US statelevel data to test this prediction and show that the states with high personal income and high tax rates are using tax amnesties more frequently. We also show that the self-fulfilling characteristic of tax amnesties may lead to sub-optimal outcomes under lack of commitment. In the third chapter, we present a theoretical framework to explain the recurring nature of tax amnesties. We construct a model with the strategic interaction of a government and a mass of taxpayers. The government and the taxpayers interact repeatedly, and each interaction can result in a tax amnesty. We show that an amnesty can cause another amnesty in the near future by altering taxpayers' beliefs about unobservable government characteristics. Therefore, an economy may get into a sequence of successive tax amnesties through a reputational channel referred to as an "expectation trap." The expectation trap mechanism can explain the series of tax amnesties in some US states that rarely experienced tax amnesties in the past. Extending our baseline model shows that a recession may cause a tax amnesty, which can trigger a sequence of further amnesties that can spread into the years after recovery from the recession. In the fourth chapter, we study the sustainable intergenerational insurance schemes under lack of enforcement. The welfare improving insurance policies may not be implementable under limited enforcement since any agent can opt-out of the insurance scheme if it does not benefit her. We develop a 2-period overlapping generations model to study the impact of different government policies. A standard tax-and-transfer scheme cannot provide perfect risk-sharing. To improve intergenerational risk sharing, we first introduce money into the system. We show that introducing money into the baseline model may improve risk-sharing. We also investigate the possibility of providing further risk-sharing by using monetary policy. We show that a monetary policy rule improves welfare under a numerical example.