Abstract

How can the international community support governance at the local level in fragile states? Across three papers, this project seeks to deepen our understanding of the localized effects of foreign aid interventions in states affected by conflict or high levels of political instability through in-depth studies of the United Nations Development Programme (UNDP). In the first chapter, I conduct a large-n, cross-national study, theorizing that more projects and more funding in a small geographic area should help mitigate violence in fragile states in Sub-Saharan Africa. UNDP, as one of the largest peacebuilding organizations in the world, has the largest peacebuilding-specific budget, works in the most number of countries, and has worked with host countries for decades to help establish a stable peace. I argue that this level of commitment means that UNDP is the most likely organization to succeed in peacebuilding despite difficult circumstances. I collected a novel time-series dataset on the location and expenditures of UNDP peacebuilding projects to test whether higher expenditures and/or more projects can mitigate violence. My results suggest that an increased number of projects in an area is associated with decreased violence, and while increased project expenditures reduces violence perpetrated by non-state actors, this result is less robust. though this is not true for an increase in project expenditures. These findings emphasize that what international actors do to support fragile states can be just as or more important than how much money is spent. In the second chapter, I drill down on the importance of local presence introduced in my study on Sub-Saharan Africa. I examine a case study of a specific project by UNDP in a fragile state to ask how the international community might leverage the abilities of local governments for effective capacity building while avoiding rent-seeking in fragile states. Prior literature suggests rent-seekers can be induced to cooperate when they are closely monitored and would credibly be held accountable for poor behavior, but international actors are poorly positioned to do these tasks. I theorize that international actors can instead delegate those responsibilities to local citizens, who, under the right circumstances (including conservative budgets), can monitor elites and credibly hold them accountable. I test this theory using mixed methods by examining the Village Development Programme, a UNDP-led peacebuilding project during and after the civil war in Nepal. I quantitatively assess my theoretical expectations with a novel dataset I compiled from internal UNDP documents. Using standard statistical regressions and a Bayesian machine learning-based difference-in-differences model, I show participation increased village governments' abilities to provide goods and services. Semi-structured interviews not only confirmed that the projects were successful but provided support for my proposed mechanism and reported little rent-seeking behavior. My third chapter is a methods paper, where I propose a method for improved event coding. Event coding -- the process of extracting information into an event dataset -- is integral to many political science studies, yet most political scientists rely on off-the-shelf event data. If political scientists need event data not covered by these datasets, they must decide whether they will manually extract information themselves or, if possible, use an automatic extraction model, both of which have advantages and disadvantages. But why not use both? I present a "human in the loop" procedure where political scientists use a large language model for initial information extraction without the need for a dictionary, and human expert coders review the cases where the large language model's output is uncertain. Using an application of event coding from newspaper articles on Nepal, I show my procedure approaches gold standards with gains in time and financial efficiency. Together, this project highlights the importance of understanding local dynamics when analyzing international statebuilding interventions in fragile states. It also demonstrates the methodological difficulties of studying these local dynamics, both in terms of data availability, data quality, and the validity of assumptions required for existing methods. More methods work is needed to continue this important work. Based on my empirical findings, local actors choose violence or corrupt behavior after weighing the costs and benefits of their options, and international actors can sometimes influence those calculations. This is not necessarily through spending large sums, but instead being present and engaging with citizens to instigate a self-strengthening cycle of governance through local institutions.

Committee Chair

Jacob Montgomery

Committee Members

David Carter; Amy Pond; Carly Wayne; Yiqing Xu

Degree

Doctor of Philosophy (PhD)

Author's Department

Political Science

Author's School

Graduate School of Arts and Sciences

Document Type

Dissertation

Date of Award

5-2-2025

Language

English (en)

Author's ORCID

https://orcid.org/0000-0003-4504-7681

Download

Available for download on Wednesday, May 07, 2031

Share

COinS