How did you come to choose your research topic?

During my Master’s at the University of São Paulo, I became intrigued by a pivotal episode in 1959, when Brazilian President Juscelino Kubitschek abruptly broke off negotiations with the International Monetary Fund (IMF). Despite Brazil’s acute balance-of-payments crisis and his ambitious “fifty years of progress in five” development plan, Kubitschek refused to adopt the IMF’s policy prescriptions in exchange for financial support. His defiance was striking not only because Brazil was a founding member of the IMF and a close US ally, but also because of the broader Cold War context in which the decision took place.

What puzzled me even more was a recurring comparison made at the time by US, IMF, and Brazilian officials: Brazil versus Argentina. While Brazil struggled to finalise a few agreements with the IMF, Argentina — a wartime pariah that had been excluded from the 1944 Bretton Woods negotiations — managed to secure six IMF agreements during the same period.

This contradiction raised a fundamental question: why did two countries with such similar economic needs and geopolitical circumstances receive such different treatment from the IMF? 

The dominant literature in international political economy, heavily reliant on statistical models to identify the macroeconomic and geopolitical determinants of IMF lending patterns, offered no satisfying answer. So I asked myself: What’s missing?

I realised that the key lay not only in observable variables, but in the decision-makers themselves. This insight is especially relevant considering the IMF’s own Conditionality Guidelines, which state that the Fund should approve a member’s programme only if it perceives the member to be sufficiently committed to carrying it out. In fact, roughly 70% of the IMF’s budget is devoted to missions that gather precisely this kind of qualitative, perception-based information.

To explain the divergent experiences of Brazil and Argentina, I set out to reconstruct how policymakers understood the negotiations, interpreted their economic and geopolitical contexts, and made decisions under conditions of uncertainty and asymmetric information. What narratives were they constructing? How did their emotions, beliefs, and worldviews influence their choices?

Policymakers’ perceptions, I came to believe, might hold the missing piece of the puzzle.

Can you describe your thesis questions and your methodology for addressing them?

My thesis investigates whether and how policymakers’ perceptions influence the negotiation processes and outcomes of IMF stabilisation programmes. The central question is: Why did two countries with broadly similar political and economic contexts experience such divergent outcomes in their interactions with the IMF?

To answer this question, I use an innovative mixed-methods approach that combines large-scale text analysis with historical interpretation and econometric modelling to construct novel sentiment indices from thousands of confidential policy documents from the IMF, as well as from US, Argentine, and Brazilian governments. These sentiment indices trace how policymakers from all sides communicated their evolving views about stabilisation programmes and financial negotiations over time.

What are your major findings?

My research finds strong evidence that policymakers’ perceptions played a critical role in shaping IMF interactions with Argentina and Brazil during the period studied.

A process of “learning by doing” emerged over time. IMF officials began with rather “flat priors” about each country, but gradually gathered what I call “soft information” — perceptions about each government’s and its authorities’ trustworthiness, credibility, and capacity for policy implementation — through repeated missions and human interactions. This evolving soft information significantly influenced lending decisions, even after controlling for macroeconomic indicators and geopolitical alignment.

Through a comparative case study, my research shows that IMF lending decisions do not occur in the context of one-shot bargaining games. Instead, they unfold in a setting of incomplete information, where evolving mutual perceptions on all sides shape outcomes in ways that traditional explanations — largely grounded in economic and geopolitical observables — fail to capture.

What useful lessons could your history thesis provide for today, in terms of policy and research? 

This research contributes to multiple disciplines, such as history, economics, sociology, international political economy, and data science, by highlighting the pivotal role of policymakers’ perceptions — or “soft information” — in shaping IMF interactions with Argentina and Brazil in the early post–World War II era.

Methodologically, my research expands the historian’s toolkit by demonstrating how natural language processing (NLP) can be used to analyse large volumes of archival material. It invites historians and social scientists alike to expand the use of NLP beyond its conventional use with public sources. 

This thesis opens promising avenues of research: extending the analysis of soft information in IMF lending to more recent periods and diverse regions and using these insights to inform the design of future conditionality frameworks. More broadly, it offers a model for how international historians and social scientists can systematically examine the role of perceptions in shaping decisions, behaviour, and outcomes across a wide range of global interactions.

What are you currently doing? Are you going to continue your research?

I am currently a Senior SNSF Researcher at the University of Lausanne, where I collaborate with Professor Edoardo Altamura on the project A Matter of Life and Debt: The Lost Decade and the Rise of Neoliberalism in Latin America, 1982–1994. Building on the sentiment analysis framework developed in my thesis, I am applying these methods to IMF and World Bank programmes in Argentina, Brazil, and Mexico to investigate how evolving perceptions influenced policy decisions during one of Latin America’s most turbulent periods.

In parallel, I consult for the IMF, using similar analytical tools to help the institution better understand its historical and contemporary interactions with member countries.

Looking ahead, my research agenda centres on two key goals: first, continuing to explore how emotions, perceptions, and cognitive processes shape decision-making in both historical and present-day global governance; and second, advancing the use of NLP in historical research to open new frontiers for analysing large archival datasets.

*     *     *

Fernanda Conforto de Oliveira defended summa cum laude avec félicitations du jury her PhD thesis in International History, titled “Opening the Black Box of Financial Negotiations: The IMF, Argentina, and Brazil in the Post-war Era (1945–64)”, on 14 March 2025. Professor Nathan Sussman presided over the committee, which included Professor Rui Esteves, Thesis Supervisor, and Professor Michael D. Bordo, Department of Economics, Rutgers University, New Brunswick, New Jersey, USA.
For the excellence of her doctoral research, Dr de Oliveira is co-recipient of the 2025 Pierre du Bois Prize, awarded each year for the best PhD thesis in International History defended at the Geneva Graduate Institute.

Citation of the PhD thesis: 
de Oliveira, Fernanda Conforto. “Opening the Black Box of Financial Negotiations: The IMF, Argentina, and Brazil in the Post-war Era (1945–64.” PhD thesis, Graduate Institute of International and Development Studies, Geneva, 2025.
Access:
Members of the Geneva Graduate Institute can access the thesis via this page of the repository. Others can contact Dr de Oliveira.

Banner image by Bumble Dee / Shutterstock.com.
Interview by Nathalie Tanner, Research Office.