The cost to Brazil · study
The Brazilian crisis of 2014-2016: differences in per capita GDP trajectories since Lava Jato
- Authorship
- Helder L. Ferreira-Filho, Jefferson S. Fraga, Nelson H. Barbosa-Filho
- Institution
- National Association of Graduate Centers in Economics (ANPEC)
Academic paper (UnB, UFS, FGV EESP and IBRE) submitted to the ANPEC 2025 National Meeting; it is not a peer-reviewed publication. - Year
- 2025
- Period analyzed
- 1990–2019 and 2000–2019; shock dated to 2014
- Document
- A crise brasileira de 2014-2016: diferenças nas trajetórias de PIB per capita a partir da Lava Jato · original ↗
Method
Controle sintético sobre PIB per capita (US$ constantes de 2017), com amostra de 133 países e de países da América Latina; placebos para Chile, México e Paraguai.
Results
| Average difference in per capita GDP versus the synthetic control, 2014–2019, world sample 1990–2019 | -7 pct_pib_per_capita | source |
| Average difference in per capita GDP versus the synthetic control, 2014–2019, world sample 2000–2019 | -6 pct_pib_per_capita | source |
| Average difference, Latin America 1990–2019 | -5 pct_pib_per_capita | source |
| Average difference, Latin America 2000–2019 | -5.5 pct_pib_per_capita | source |
Study results are estimates from the authors' model, not measurements.
Assumptions
- The shock is Lava Jato 'and associated factors,' beginning in 2014.
- Countries with similar pre-shock characteristics would have a trajectory similar to Brazil's without the shock.
Limitations
- It does not separate Lava Jato from other shocks around 2014 (oil price, water and energy crisis, 2013 protests, economic policy); the authors acknowledge the limitation.
- Synthetic Brazil composed of a few countries, sometimes very different from one another (Ecuador, Uruguay, Congo, etc.).
- The results measure what fell below the counterfactual, not why.
Divergences and other studies
It measures per capita GDP accumulated over six years, a quantity different from the consultancies' estimates (GDP points in 2015) and from DIEESE's (GDP accumulated 2014–2017); do not compare directly. The article's own literature review records synthetic control studies (Balassiano 2018; Pessoa 2017) that attribute 40% of the slowdown to international factors and 3.1 points per year of difference.
Extracted evidence
A paper submitted to ANPEC 2025 (Ferreira-Filho, Fraga and Barbosa-Filho) applied synthetic control to per capita GDP and found that, from 2014 to 2019, Brazil ended up 5% to 7% below the counterfactual (four exercises: 7% and 6% with 133 countries, 1990–2019 and 2000–2019; 5% and 5.5% with Latin American countries), with placebos (Chile, Mexico, Paraguay) showing no similar divergence.
Source: A crise brasileira de 2014-2016: diferenças nas trajetórias de PIB per capita a partir da Lava Jato — ANPEC (2025 National Economics Meeting, submitted paper), 2025 · Section 4, Table 2, p. 13; Conclusions, p. 14 · open original ↗
Study (primary) · full text examined · accessed on Oct. 24, 2026
Limitations: The 'shock' is Lava Jato 'and associated factors,' dated to 2014: the method does not separate the operation from the fall in oil prices, the political crisis, economic policy and the 2013 protests; the authors themselves list as a limitation not accounting for other shocks close to the date. Synthetic Brazil is formed by a few countries (for example, Ecuador and Uruguay), with variable pre-event fit. Paper submitted to a conference, with no known peer review. One of the authors (Nelson H. Barbosa-Filho, FGV/IBRE) is identified by Folha, in 2021, as a former finance minister; in the same report he observed that part of the losses may reflect misallocation of resources caused by corruption.
The literature review of the ANPEC 2025 paper summarizes studies that attribute the slowdown to various factors: Borges (2017) estimates 40% to 60% of the 2012–2016 slowdown due to exogenous factors (international and domestic), about 30% due to underestimation of GDP and 10% to 30% due to policy errors; Balassiano (2018), with synthetic control of 14 countries, arrives at about 40% of the 2014–2016 slowdown due to international factors; Pessoa (2017) finds an average difference of 3.1 percentage points per year between actual and synthetic Brazil; Matos (2016) estimates about 30% due to external factors.
Source: A crise brasileira de 2014-2016: diferenças nas trajetórias de PIB per capita a partir da Lava Jato — ANPEC (2025 National Economics Meeting, submitted paper), 2025 · Section 2 (background and literature), pp. 5–7 · open original ↗
Study (primary) · full text examined · accessed on Oct. 24, 2026
Limitations: Summary made by third parties; the original studies (Borges 2017; Balassiano 2018; Pessoa 2017; Matos 2016) were not opened in this version of the collection. The results measure external vs. internal factors, not the effect of Lava Jato.