How Cost-Efficient Is USAID? A Data-Driven Look at U.S. Foreign Aid Agencies
while measuring actual cost-efficiency
Introduction: USAID’s Role and Foreign Aid Under Scrutiny
Since its creation in 1961, the United States Agency for International Development (USAID) has been the nation’s primary foreign aid agency . It administers tens of billions of dollars in humanitarian and development programs worldwide each year, making it a pillar of U.S. “soft power.” Proponents credit USAID with saving lives and promoting stability abroad, which in turn bolsters U.S. national security and global goodwill. In fiscal year 2023 alone, USAID distributed nearly $43.8 billion in aid – about three out of every five U.S. foreign assistance dollars. By comparison, the State Department disbursed about $21.3 billion (almost 30% of the total), with smaller shares handled by the Department of Defense (DoD), the Millennium Challenge Corporation (MCC), and other agencies. Despite the relatively modest slice of the federal budget that foreign aid represents (roughly 1% of spending), USAID and U.S. aid programs have come under intense political scrutiny. In early 2025, the Trump administration froze most foreign assistance for 90 days pending a sweeping review of “efficiencies” and alignment with U.S. interests.
Musk and his DOGE set their sights on foreign aid early in the administration. Both Musk and lawmakers in this camp argue that foreign aid is bloated and ineffective, and some openly called for slashing USAID’s budget or folding its functions into the State Department (pewresearch.org). Against this charged backdrop, it’s important to ask: How cost-efficient is USAID in delivering foreign assistance compared to other agencies that also provide foreign aid?
In this post, I take a data-driven approach to evaluate the cost-efficiency of USAID relative to other U.S. foreign aid channels, namely the DoD, State Department, and MCC. Before DOGE decided to gut all access to any USAID spending data (I believe it is back online) I compiled a dataset of spending on U.S. foreign assistance projects – which includes a variable indicating whether each project was implemented by USAID or another agency. I was saving the data for another paper I had in mind but, given the circumstances, I decided to examine whether USAID was more (or less) cost-efficient than other USG organizations.
The goal of this post is to present preliminary results on whether USAID truly lags behind on efficiency or if it delivers results on par with (or better than) other agencies. The analyses draw on available US government spending data and foreign aid outputs to offer insights on a debate that has been mainly driven by anecdotes and ideology.
What is a Stochastic Frontier Cost Model, and Why is it probably better than using anecdotal evidence to eliminate foreign aid ?
When policymakers and critics evaluate government efficiency—especially regarding foreign aid—it's easy to cherry-pick examples that support a particular narrative. This often happens when the Department of Government Efficiency (DOGE) and Congressional committees scrutinize USAID, focusing on anecdotal evidence or individual problematic grants. However, what we have seen is blatant cherry-picking to make bold—and often inaccurate—claims to justify the gutting of a critical organization for international and domestic stability. Also, such selective storytelling misses the bigger picture: how does an agency perform overall, compared systematically to others doing similar work?
This is exactly where a stochastic frontier cost model comes into play. Rather than cherry-picking single examples, the stochastic frontier approach rigorously examines all data points simultaneously to determine how close each agency's spending is to a theoretical "best practice"—the cost frontier.
How Does it Work?
In a stochastic frontier cost model:
Considering available resources and external conditions, we estimate a "cost frontier"—the lowest possible cost that can achieve a given outcome. This frontier represents maximum efficiency.
The cost of every observed program is then compared against this frontier. Deviations from the frontier reflect inefficiency—the gap between the actual spending and the lowest possible achievable spending given the outputs delivered.
The model explicitly separates inefficiency (systematic overspending or waste) from random noise (unexpected events or measurement errors) by splitting the error term into two distinct parts:
A random, normally-distributed error capturing factors outside an agency's control (like unforeseen events or measurement error).
An inefficiency term, usually modeled as a truncated-normal or half-normal distribution, explicitly capturing cost inefficiency—avoidable waste or excess spending that an agency could theoretically control or reduce.
This approach is powerful because it recognizes that not all variation in cost is due to inefficiency. Instead, it isolates true inefficiency from random fluctuations, allowing for a much fairer and clearer comparison of agency performance.
I’m all for effective foreign aid reform and improving its efficiency. However, doing that requires careful and thorough analysis. DOGE’s current approach, however, relies on individual anecdotes or aggregate numbers— and heavy ideological views— without examining the relationship between inputs, costs, and outputs.
Is (or was) USAID more cost-efficient?
A significant challenge in assessing the cost-efficiency of foreign aid is accurately defining and measuring "outputs" or outcomes. Unlike businesses, for example, where outputs such as units produced or sales revenue are easily quantifiable, foreign aid produces less tangible outcomes like improved governance, reduced violence, better health indicators, or stronger institutions. How do we place a meaningful numerical value on stability, governance quality, or safety? First, I focused on some of the main results these programs are trying to achieve, namely higher stability, lower violence, lower corruption, etc. A conventional approach in efficiency analysis is to use the reciprocal of an indicator to capture—or at least approximate— the work produced by an organization. Specifically, I used the following indicators for the analysis1:
Reciprocal homicide rates: This measure inversely captures security outcomes—lower homicide rates indicate greater stability, which theoretically should reduce costs for aid programs operating in safer environments.
Corruption control indices: These provide insight into governance quality. Better governance should facilitate smoother aid operations and lower associated transaction costs.
Conflict intensity scores: Similar logic applies here—less conflict intensity should correlate with more manageable operational contexts and thus lower costs.
Although some may argue that these outcome variables are unconventional, they do capture the overarching outcomes associated with development interventions.
In this sense, rather than relying on anecdotal or cherry-picked cases, as some critics of USAID do, I try to assess aid efficiency against quantifiable benchmarks that reflect conditions on the ground and offer insights on whether USAID is relatively more cost-efficient.
Key Results
Table 1 below displays the results of the stochastic frontier analysis cost model. It is important to emphasize the following findings. First, the USAID dummy— which directly compares USAID programs with those run by other agencies (such as the Department of Defense, State Department, and MCC)—is negative and satistically significant (p<.10). This indicates that USAID may operate closer to the cost frontier than the average of other agencies. Thus, one could reasonably argue that USAID generally achieves similar outcomes at lower costs compared to other foreign aid investments by other U.S. government agencies.
The remaining output measures in the model also underscore the importance of investments in foreign aid. In particular:
Better governance and security conditions lead to more cost-efficiency, which I argue, aligns with common sense. When USAID and other agencies operate in environments with lower corruption and lower violence, they can deliver results more cheaply and efficiently.
Why Do These Results Matter?
The findings presented speak to several dimensions of the foreign aid discussion. First, the findings directly challenge the narrative advanced by DOGE and critics who, without conducting efficiency analyses, claim USAID is inefficient. The evidence suggests otherwise, showing that USAID performs more cost-efficiently than its peers, such as DoD and other US agencies that provide foreign assistance. Second, if USAID is more cost-efficient, reducing its budget or merging it with, say, the State Department, could disrupt this efficiency ( I am being kind here). This is critical given USAID's role in promoting stability and security abroad, which also supports U.S. national security interests. For instance, sub-Saharan Africa, receiving over $6.5 billion in humanitarian aid last year, could suffer significantly from any cuts, impacting programs like HIV support. Finally, this contrasts with DOGE's approach, which relies on anecdotes, and further underscores the need for systematic evaluations. This is supported by recent USAID efforts, such as the 2024 Position Paper on Cost-Effectiveness (Committing to Cost-Effectiveness: USAID's New Effort to Benchmark for Greater Impact). There should be ongoing discussions to reform foreign aid and improve its efficiency without indiscriminately cutting programs.
These data were obtained from the Quality of Government Dataset at the University of Gothenburg. Link: https://www.gu.se/en/quality-government/qog-data/data-downloads