2021 Shortlist

120 years of darkness: shedding light on government pensions in Brazil

Country/area: Brazil

Organisation: Fiquem Sabendo

Organisation size: Small

Publication date: 13 Jan 2020

Credit: Maria Vitória Ramos, Bruno Morassutti, Luiz Fernando Toledo, Léo Arcoverde, Fernando Barbalho, Álvaro Justen

Project description:

For over a century Brazilian tax payers have given a blank check for the government to use on pensions for public servants with no oversight. For three years Fiquem Sabendo battled in the Supreme Audit Institution of Brazil,  pressuring for access to historical records. In 2020 we won, freeing 26 years worth of data about payments. We cleaned, analysed and uploaded the data online, identified issues and corrected the official database by collaborating with the  Ministry of Economy and then organized a pool of journalists to produce stories. Finally, we gave free access to our tool for everyone. 

Impact reached:

This was possibly the largest database ever obtained based on Freedom of Information in Brazil. The data on individual monthly pension payments to civilians from 1994 to 2020 consists of 100 million lines spread across 27 different files. It is also – possibly –  the most important in terms of public money disclosure. We are talking about U$ 88 billion in pensions for over 400.000 public servants’ family members, through 95 million individual payments made in 26 years. In the dataset we found lifetime pensions established in 1900 and payments that reach up to R$ 30,000 monthly. Before that, it was impossible to hold the government nor beneficiaries accountable because their names and pensions were unknown to the public.

 To take these historical numbers to the broader public, Fiquem Sabendo coordinated a collaborative task force with four journalists to produce stories from the dataset. Lucio Vaz (Gazeta do Povo), Eduardo Barretto (Época), Bruno Fonseca (Agência Pública), and Taís Seibt (Fiquem Sabendo/Yahoo) worked for over a month with our team to find scoops. Agência Pública, for example, found that hundreds of officers accused of torture during the dictatorship have been receiving lifetime pensions from the government. 

Over a year since the data was released by our small independent team, it is still being used by many major news outlets (Google “Fiquem Sabendo” “pensionistas”). Even president Bolsonaro tried to weigh in the situation by falsely claiming on social media he was the one to order the release of the dataset. The House of Representatives leader condemned the pensions.

The government was positively impacted by our investigation. We will explain later how our work changed the official records by almost a billion dollars in net value by identifying structural mistakes on the dataset and helping the Ministry of Economy to fix it. 

Techniques/technologies used:

We managed to transform the 60 GB dataset from an inaccessible archive of separated files into a dynamic and – most importantly – useful online dataset for reporters, researchers, and citizens. 

On Shinyapps, the application structured by Fernando using R, anyone is able to explore filters, timelines and download smaller pieces for further exploration on Excel/Sheets. A very important aspect here is the monetary corrections applied to update the numbers. Through this tool, anyone can see both the original values from the official dataset and the new values considering two decades of inflation and a change of currency. The platform is still online and open to the public.

On Metabase, structured by Álvaro Justen using Python, it was possible to create dashboards of interactive graphs, make calculations, pivot tables, and follow a single beneficiary’s complete path over the years. Because this is a tool that costs based on the usage volume, we gave exclusive access for journalists to deeper explore the data.   

What was the hardest part of this project?

Overcoming public data censorship Fiquem Sabendo started this fight back in 2017 and only after three years in the Supreme Audit Institution of Brazil, we won. The decision, unanimous and favorable to the agency, guaranteed the publicity of payments made to pensioners of the federal government for the first time in history. In January of 2020 we got two months worth of data and – finally – in July we released 26 years of payments for accountability. This was a result of a long collaboration between journalism and law. Despite the huge achievement, we are still in a battle for the full release of the information! Over a year after the tribunal’s decision, the government has not yet published the data on payments made to relatives of military personnel and secret agents, nor to relatives of Central Bank employees. Fiquem Sabendo has already filed two petitions for this information to be made public and as soon as it is we will repeat the same process. Identifying and correcting mistakes in the official dataset As we analysed the dataset we noticed strange entries. Because we had built two different tools, in different languages ​​(R and Python), by two different people, we were sure that the problems detected on our platforms were on the official data published by the Ministry of Economy. We then organized the probable errors and inconsistencies found in three spreadsheets: single payments over R$ 1 million, R$ 100,000 and empty cells. For over a month we went back and forth with the Human Resources Management Secretariat to show and correct the problems. How impactful were those mistakes in the public budget? Almost a billion dollars in net value. Because we collaborated with the technical department responsible for the information, our reporting was able to correct the original governmental

What can others learn from this project?

Collaboration with other areas, not just other newsrooms

Having a lawyer as a cofounder working to support journalists showed us how important it is to have a team from diverse areas. Our reporting would not have existed if it wasn’t for the whole legal battle that preceded it and the court’s decision would not have invoked such public appeal if it had not been appropriately presented to society. 

Open for everyone

Leaks and special sources are part of journalism, but we also need to fight to make things actually public. It is a structural part of our job to make public data available not just to do a scoop, but to create real accountability by allowing anyone to check the full data – available online. By using FOIA the reporting is not just about that one piece anymore, it is a pathway for others to push transparency even further. Documents obtained through the legal due process become a precedent for other documents, agencies and for state and city level decisions. The importance of this project does not end on itself, it can now be used to open pension payments all over the country, it can base requests for other parts of the government, etc. 

Citizens are potential journalists waiting to act

They have the will. If we share the  tools, they can help us. It was through Twitter DMs of family members that didn’t find their own relatives on the dataset that showed us the information first released by the government was not complete. “Normal people” helped us identify that the first dataset published by the Ministry of Economy did not include beneficiaries related to military personal, secret agents and the Central Bank. And that’s what fueled our second petition to the Supreme Audit Institution of Brazil.

Project links:

fiquemsabendo.substack.com/p/aps-denncia-da-fiquem-sabendo-governo-17d

fabdev.shinyapps.io/graphs_on_demand/

fiquemsabendo.com.br/transparencia/pensionistas-e-aposentados-servidores-inativos/

br.noticias.yahoo.com/governo-federal-pagou-r-384-bilhoes-a-pensionistas-desde-1994-070059010.html

apublica.org/2020/01/as-pensoes-vitalicias-dos-acusados-de-crimes-na-ditadura/

piaui.folha.uol.com.br/pensao-a-brasileira/

politica.estadao.com.br/noticias/geral,governo-paga-pensao-a-52-mil-filhas-solteiras-de-ex-servidores-do-executivo,70003185132