In 28 years, Bolsonaro clan has named 102 people with family ties in their offices

Category: Best data-driven reporting (small and large newsrooms)

Country/area: Brazil

Organisation: Jornal O Globo, Revista Época

Organisation size: Big

Publication date: 8 Apr 2019

Credit: Juliana Dal Piva, Juliana Castro, Rayanderson Guerra, Pedro Capetti, Marlen Couto, Bernardo Mello, João Paulo Saconi, Daniel Lima, Gabriel Godoy, Fernando Lopes, Ana Luiza Costa

Project description:

After an investigation into 28-year public data, we found that President Jair Bolsonaro and his three parliamentary sons have had 286 advisers since 1991. In total, 102 had a family relationship. From the Bolsonaro family alone, 22 named relatives were discovered. We also found that of this total, 37 were phantom employees who received public money without working for years. They were nannies, housewives, retirees. The revelations fueled official investigations by the authorities into two of Bolsonaro’s sons who raised suspicions that they were getting their salaries in a big money-laundering scheme.

Impact reached:

The report has given another dimension in the story that is known in Brazil as the “Queiroz Case”, the main event of corruption involving the Bolsonaro family. Until that time, the Rio de Janeiro Public Prosecutor’s Office was investigating nine advisers of Flavio, Jair Bolsonaro’s eldest son, and many suspicions that Flavio was charging ilegally his employees part of their salaries. The report has shown this modus operandi was biggest and widespread in the Bolsonaro family offices.

The publication has been divided between the two vehicles of the group: newspaper O Globo and magazine Época. The initial presentation of data of the 102 named advisors came out at O Globo, and the magazine article presented the impact of R$ 65.2 million on public coffers, showing to the public the dimension of this crime that, even though is common in Brazil, is seen as a minor corruption.

With the data revealed, authorities have deepened the investigation into Flavio and found that 10 of his relatives used to withdraw up to 99% of their salaries every month, always on the payday of the Legislative House. At the same time, came the suspicion that this money was laundered with real estate purchases.

Prosecutors also opened an official investigation into Carlos, another Bolsonaro’s son, for of the same suspicions, and the basis for the opening was the allegations published in the reports. Even though he was a city councilor for the city of Rio de Janeiro, people living in other towns and states advised him, which is against the law. Even an ex-wife of the president has become investigated.

Bolsonaro threatened the press the day after the reports were published. He reduced one of the sources of funding for media companies. After this, he also began to make changes in investigation institutions. 

Techniques/technologies used:

With the immense amount of information investigated in three Legislative Houses and official newspapers, the reporting and infographics team have built a unique database that was unavailable to the public – much data of three decades was documented only on paper and was lost in the archives. This material has become relevant not only for gathering information about the President’s family but also for his 28 years of public life.
Then, the team decided to present and draw these family links in a digital environment showing the kinship relations between the advisors in the four offices. With the aid of algorithms of a programming language and statistics (R), the connections were organized and compiled in a format that allowed the construction of an interactive network diagram (made with the D3 package, in JavaScript / HTML / CSS) that the reader can consult at ease. A real database was made available to anyone who wanted to see, with information about the profession, salaries, and kinship of advisors.

An interface (also with React / JavaScript / HTML / CSS) was designed for individual consultation of each advisor, optimized for smartphones. In this way, we made public a variety of information that should appear on websites of Legislative Houses, but were not available, in violation of Brazil’s law on access to information. The infographic also allowed the readers to access the report as they pleased.

We also recorded a podcast explaining all the process of the reporting and using all of the recordings we made in interviews with these ghost employees. This way the public could hear some of the confessions that this people made when they were asked about the reality that they did not worked for Bolsonaro’s family.

What was the hardest part of this project?

We have requested Information using a properly law in Brazil for acess of public data to be able to know the list of all Jair Bolsonaro staff in the House of Representatives since 1991, his first term. In those requests we also asked the periods in which these people supposedly worked, positions and salaries. All information is public by law in Brazil. The same request was made to the Rio de Janeiro Legislative Assembly and to the Rio de Janeiro City Council – where the president’s son were elected.

In the House of Representatives and the Legislative Assembly the requests were fulfilled within the deadline of the law, which is 20 days. Many documents were delivered in print or PDF format. This way, about a thousand pages of information had to be converted into a spreadsheet so that the data could be further analyzed. The gathering of all information was done by the entire team of reporters.

The Rio de Janeiro City Council ignored the requests for data. Given the lack of response, 3 reporters decided to do the survey personally because this data was at least registered in printed papers on the City Council archive. The only online information available is from 2017. We needed data from back 2001 until 2018. So we manually searched all the papers of 18 years to find all the names of the employees, positions and periods.

After all the data was analyzed extensive investigated 37 people to prove that they were ghost workers. We made several trips to other cities, consulted several databases and interviewed several people. Most employees declined to interview, many were afraid of police investigated for homicides who had relationships with Bolsonaro family employees. Three people, however, even confessed that they never worked.

What can others learn from this project?

To Brazil we created an important database that did not exist. This work was done in a concentrated manner over 3 months, but we began to investigate almost a year earlier.

This data was scattered in three legislative houses with lack of transparency and, in some cases, were even lost in the archives of these institutions. With the investigation we made this information available online, what should have been done from the beginning by the authorities themselves.

The interface created to provide queries to the database allowed the reader to see the story in different ways. It is posible to search for a unique person or for an entire family. The readers can also see the connections these people have with Bolsonaro’s family. There are 286 online individual files and the advisers who are being investigated are detailed on these. 

This was a huge learning process because it enables multiple readings and this is something that reporter teams need to start thinking about when working with large databases. The general data is often not as interesting as the unique stories inside them. 

This was a huge learning because it enables multiple readings and this is something that reporter teams need to start thinking about when working large databases. The general data is often not as interesting as the unique stories.

We also combined a large investigation to have the data only  to create the very data base we needed for the special reporting. Only after that we went to the street to do the research on the ghost workers.

I think this work shows that data-driven research will not always be from a ready-made database, but one that needs to be built with a lot of persistence and only then can software be used for data analysis work.

Project links: