Parts of this dissertation have been presented at various conferences and seminars. For their suggestions and comments, I would like to thank Kenneth Binmore, Francesca Carapella, Roozbeh Hosseini, Facundo Piguillem, Evsen ? T? rkay, and the members of the Chari-Jones Workshop, as well as the particiu pants at the III Guanajuato Workshop for Young Economists, the Third Annual Graduate Student Conference at Washington University in St.Louis, the 6th Midwest International Economic Development Conference, the 2009 European School on New Institutional Economics, and the 13th Annual Conference of the International Society for New Institutional Economics.
The chapter titled “Self-selection of Politicians and Corruption” is written with Evsen T? rkay. ? u i The Department of Economics at the University of Minnesota is blessed with the best staff! They have not only made my life so much easier by taking care of all the bureaucratic processes related to my studies, but also offered me their moral support and encouragement on a daily basis.
I especially thank Catherine Bach and Kara Kersteter for all they have done for me from the time I took preliminary exams to the time I was on the job market. I would like to thank Dr. Simran Sahi for her patience and understanding during my training as an instructor. I learned so much about teaching from her! I am grateful for her moral support as well as professional mentorship. I would like to thank the Research Department at the Federal Reserve Bank of Minneapolis for hosting me in the past four years. The work environment they provided helped me be more productive. I thank very much to Dr.
Warren Weber for allowing me to have a work space at the Fed and staff members Barbara Drucker, Vicki Reupke, and Wendy Stier for their kindness. As anyone who has ever written a doctoral dissertation would agree, continuous moral support by family and friends is very important during this long and exhausting process. I consider myself very lucky to have many loving and supportive people in my life, who provided me with an unbelievable amount of encouragement and support. First of all, I would like to thank my parents Nevin Tarhan and R. Engin Tarhan for all they have done for me from the moment I i was born. Without their continuous support, both material and emotional, and unconditional belief in me this dissertation could not have been written.
I show that a self-interested government sets a higher public-to-private-capital ratio than a benevolent one in order to increase the before-tax returns to private investment and hence increase tax revenues that can be expropriated. However, after-tax returns to private investment are lower and hence the growth rate is lower. Another result is that self-interested governments choose a high level of non-productive public investment, which provides a channel for the government to expropriate tax revenues for its private gain, thereby in? ating total public investment.
In the model I assume public investment to be ? nanced through income taxes. Collecting taxes and deciding how to use the tax revenues give the government an opportunity to engage in corrupt activities for its own bene? t. Using the model, I study the choices of the government and the behavior of consumers as a response to government policies, all depending on how benevolent the government is. In the model the government is assumed to maximize a weighted average of consumers’ welfare and its own welfare coming from expropriated tax revenues. The weight on consumers’ welfare determines how benevolent the government is.
If the weight on consumers’ welfare is zero, then the government is totally self-interested, and if the weight is one then the government is totally benevolent. The weight can be any number between 0 and 1, implying that the government can be partially benevolent. I show when the government is selfinterested, the amount of productive public investment is low but the amount of expropriated tax revenues is high. The government is assumed to be constrained by a period-by-period budget, which implies an upper bound on total embezzlement by the government at any period.
This results in a dilemma for the corrupt politicians: they can either steal as much as they can at any period, leaving only a small amount of funds for the ? nancing of the public capital, or they can invest in public capital so as to increase the productivity of private capital, and hence income, in the future. Increased income implies higher income tax revenues and more funds to embezzle in the future. Therefore, each type of government chooses an optimal growth rate through its policies that balances the cost of deferring expropriation 2 of funds today and the bene? t of increased tax revenues that can be embezzled in the future.
This optimal growth rate is determined by the public-to-private capital ratio. I argue that a self-interested government chooses a higher publicto-private-capital ratio than a benevolent government and that this results in lower economic growth in the long run. The model predicts low productive public investment and low growth in countries with self-interested governments. When testing the predictions of the model against data, benevolence of a government will be thought of as the degree of lack of corruption in that country. Hence, a self-interested government in the model will be a counterpart of a highly corrupt government in the data.
While the model distinguishes between productive public investment and expropriated tax revenues, it is hard to do so in the data. Expropriated tax revenues are recorded as part of government budget and affect several entries in the government budget. However, authors such as Tanzi and Davoodi (1997) and Keefer and Knack (2007) claim that most of the corrupt activities of governments are recorded as public investment1 . Treating expropriated tax revenues as part of public investment in accordance with these studies, the model predicts that high levels of total public investment would be observed in countries with high corruption.
To the best of my knowledge, this paper is the ? rst attempt to explain the interrelationship between political corruption, public investment, and economic growth through a model that analyzes the behavior of different types of government. Haque and Kneller (2008) undertake an empirical study to see the effects of corruption on public investment and economic growth. They ? nd that corrup1 See next section for a more detailed discussion. 3 tion raises the level of public investment but lowers the returns to it, making it ineffective in promoting economic growth, which is consistent with the results of my model.
Background and Related Literature The effect of public investment on growth has been debated extensively in the literature. Starting with Barro (1990), many researchers have tried to capture the effect of public investment on growth; however, a consensus on the empirical evidence has never been reached. There are studies claiming that public investment is not important for economic growth (e. g. Easterly and Rebelo (1993)) while others maintain that public investment has a substantial positive effect on growth (e. g. Aschauer (1989)).
There are yet other papers which assert that only certain types of public investment are productive and that the effect of these on growth are different from the effect of non-productive public investment. For example, Devarajan, Swaroop, and Zou (1996) ? nd that current expenditure has a positive effect on economic growth whereas capital spending of governments has a negative relationship on growth. They argue that developing countries have over-invested in public capital at the expense of current spending. The link between corruption and public investment has been explored mainly empirically.
Tanzi and Davoodi (1997), for example, maintain that corrupt governments choose a higher public investment share of aggregate income. They claim that political corruption is often tied to capital projects. This is because the decisions regarding the budget and composition of capital are highly discretionary. Lack of competition in undertaking big capital projects and the dif? 4 culty in assessing the real cost and value of these projects make them a tool for corruption. The authors also argue that corruption reduces the productivity of public capital.