The political economics of earmarked taxation

This dissertation examines the decision by a government to use earmarked taxation. This dissertation is concerned with the public finance definition of earmarking. In public finance, an earmark is a requirement that all or a portion of a certain source of revenue, such as a particular tax, be devoted to a specific public expenditure. Without a change in policy the earmarked revenue will continue to be dedicated to its purpose on into the future. The first essay of the dissertation demonstrates in a world absent of frictions, a legislature will choose to earmark all of its revenue. The second essay shows how political uncertainty and party politics create frictions that prevent earmarking all the time and empirically shows that when the frictions lifted US states earmark a higher percentage of their revenue.;The first essay develops a model of legislative spending in which revenues can be spent through earmarks or a general fund. Legislative choice is modeled as a Baron and Ferejohn style legislative bargaining game. The novel approach is to model the bargaining process as a two stage game reflecting the reality that earmarked spending precedes general fund appropriations. This reality is what drives the result that all revenue is spent by way of earmarking leaving no revenue in the general fund. This model is the first to show that earmarking can occur endogenously as equilibrium behavior in a legislative bargaining game when frictions are absent.;The second essay provides a theory of earmarking based on the relative power of state legislature and the governor along with party politics. These two entities interact in a game with party specific preferences and uncertainty about the future. The politically powerful use earmarking as a way to resolve uncertainty and constrain the future government. The model predicts that earmarked policies will be implemented when the government is controlled by one party or when legislative power is so concentrated in one party's control that it may unilaterally overturn a governor's veto. These predictions are empirically tested using spatial econometric techniques on a panel of US state data.