Economics of Public Procurement: Analyses of Corruption and Collusion Using Structural Modeling

Abstract

Public procurement is one of the most critical aspects of government spending, accounting for roughly one-third of all government spending or 12% of GDP. My dissertation consists of three chapters that examine the economics of public procurement. This dissertation applies modern economic tools such as structural modeling, causal inference, and machine learning to study the most important issues of this market. In Chapter 1, I study the behavior of government agencies and its implication for the optimal design of public procurement. I develop a model that measures the incentives of government agencies and quantifies inefficiencies. Using data from the Czech Republic, I show that government agencies behave inefficiently due to corruption, under-investment into the quality of projects, and lack of information. A better market design that takes these inefficiencies into account could significantly lower government expenditures. In Chapter 2, we study a particular channel of inefficient behavior- favoritism towards politically connected firms. We create an extensive database of links between political parties and companies and show that politically connected firms deliver a large share of government contracts. Such connections then lead to unequivocally bad procurement outcomes. Prices rise, and quality does not. We show that increased monitoring likely mitigates this issue. In Chapter 3, we develop a novel tool how to detect collusion in public procurement. Using data from a new E-procurement platform in Ukraine, we use repeated interactions of companies to detect networks of colluding firms. We verify on a large sample of prosecuted companies that our novel approach is very good at detecting companies that were indeed prosecuted for collusion. However, we also show that authorities never detected a large number of collusive rings.