Bureaucracy at the Border: The Fragmentation of US Foreign Aid

Research shows that foreign aid is most effective when it is centralized and streamlined, but aid advocates complain that countries waste billions of dollars on redundant and often counterproductive foreign aid tools. Is foreign aid policy really more fragmented than ever? And how can we explain changes in fragmentation over time?

Fragmentation is a byproduct of bargaining and vote-buying in Congress. Legislators differ in their preferences over how and where foreign aid is spent. Just like the preferences of legislators themselves, foreign aid bureaucracies differ in the way they spend foreign aid money, the interests they represent, and the oversight they face from political principals. Therefore, different legislators' preferences correspond to different agencies.

When congressional agenda-setters want to pass a foreign aid bill, they might struggle to ensure that fifty percent of legislators will vote in favor of it. One solution is to appease a certain subset of influential legislators by shifting foreign aid funds from the largest, most general agency--USAID--to smaller agencies that better represent those legislators' specific interests and preferences. This attempt by agenda-setters to buy votes leads to the unintended fragmentation of foreign aid policy.

To predict the situations under which agenda-setters might be forced to fragment funds in order to buy votes, I build a theory that complements previous work on vote-buying in Congress. Based on the results of a spatial model, I hypothesize that aid fragmentation should be at its highest when the majority party is heterogeneous and parties are divided. In line with previous literature, I also hypothesize that the most likely targets of vote-buying are moderate legislators.

In an extension to the model, I also consider substitutability in vote-buying tools. I find that the presence of divided party government changes the calculus for congressional agenda-setters. Finding it counter-productive to convince moderates when a bill specifies funds and constrains the executive, agenda-setters will be more likely to use fragmentation as a vote-buying tool under divided party government.


To test the hypotheses derived from the model, I built a novel agency-level dataset of Foreign Operations Appropriations bills from 1961 to 2015. By reading these bills and recording their contents, I was able to determine the total level of spending for every line item in the foreign aid bill and aggregate these dollar amounts by agency. I then calculated the Herfindahl Index of funding between aid agencies for each year in the dataset. I also recorded the percent of each agency's funding that was specified for a given purpose or project and counted the number of general provisions for each budget year. This provided me with measures for three different vote-buying tools. I then tested the longitudinal hypotheses outlined above on a year-level dataset using a variety of statistical models for robustness. I also recorded legislator votes on each aid bill in order to test the hypothesis of majority-party moderates and vote-buying.

Quantitative Data

By conducting nearly 50 elite interviews over a six-month period in Washington, DC, along with using transcripts from congressional hearings and debates, I traced the mechanisms leading to the creation of two separate pieces of foreign aid legislation. I found that, in line with the theory, the pet projects of a small group of moderate majority-party legislators led to the creation of fragmented aid bills.

Qualitative Cases


David Lake (Co-Chair)

Christina Schneider (Co-Chair)

Lawrence Broz

Sam Kernell

Steph Haggard

Lauren Prather