Experimental Evidence of the Effectiveness of Mechanisms Designed to Increase Tax Compliance

Low tax revenues significantly limit governments' ability to provide essential public services. Despite the gravity of this problem, few systematic attempts have been made to design and evaluate feasible, scalable ways to enhance revenue collection and public sector performance. Through four related but distinct studies, this project will fill this gap by examining the impact of incentive policies on public employee performance, as well as the roles of citizen engagement and tax collectors' ability on revenue collection. Given the critical importance of these issues and the embedded and sustained nature of the researcher-policymaker partnerships, the results will provide practical guidance on increasing tax revenues to a large audience of policymakers worldwide.

While there has been significant work done on the optimal design of statutory taxes (e.g., Pomeranz 2014, Kleven and Waseem 2013), there has been less focus on increasing revenue and enforcement within a given system of taxation. This project seeks to explore enforcement and compliance issues contributing to the low collection of urban property taxes and examine the effectiveness of schemes designed to address those issues. The project consists of four studies. The first is a new randomized trial that addresses tax morale and the willingness of citizens to pay taxes. This component seeks to strengthen the link between tax payment and service provision by eliciting taxpayers' service preferences and sharing this information with their local governments. Localities will be randomly assigned to either receive the information from this preference elicitation or not, and those that receive it will be randomly assigned to allocate a portion of their budgets based on taxpayer preferences or to spend as normal. Tracking revenues, allocations, service provision, and citizen beliefs will provide a clean empirical estimate of the impact of these schemes on taxpayer attitudes and actual revenues. The next two studies focus on incentivizing tax collector performance. In the second study, tax collectors were randomly assigned to compete for preferable job postings based on their performance, which was measured by the increase in recovery or the tax base. This design allows for clean testing of the impact of these non-financial incentives on revenues and collector behavior, which will be analyzed during the course of this project. The third study will evaluate the long-term impacts of a previous study, in which tax collectors were randomly assigned to receive performance pay for two years based on revenue collection, audits, and/or subjective ratings. This performance pay ended two years ago, and the current study will determine whether this past period of performance pay had a persistent effect on tax collection' information that can help design the optimal duration for and interval between such schemes. The final study will analyze the impact of individual inspectors' characteristics on revenues by using the overlap of different collectors in the same region to separate individual from regional effects.