Download file
Download file
Download file
Download file
Download file
Download file
6 files

Central Bank of Brazil data of foreign capital transfers, 2000-2011

Download all (49.2 MB)
posted on 2018-08-13, 06:48 authored by Alice Dauriach, Emma Sundström, Beatrice Crona, Victor Galaz
This data set is a subset of the "Records of foreign capital" (Registros de capitais estrangeiros", RCE) published by the Central Bank of Brazil (CBB) on their website.

The data set consists of three data files and three corresponding metadata files. All files are in openly accessible .csv or .txt formats. See detailed outline below for data contained in each. Data files contain transaction-specific data such as unique identifier, currency, cancelled status and amount. Metadata files outline variables in the corresponding data file.

RCE_Unclean_full_dataset.csv - all transactions published to the Central Bank website from the four main categories outlined below
RCE_Unclean_cancelled_dataset.csv - data extracted from the RCE_Unclean_full_dataset.csv where transactions were registered then cancelled
RCE_Clean_selection_dataset.csv - transaction data extracted from RCE_Unclean_full_dataset.csv and RCE_Unclean_cancelled_dataset.csv for the nine companies and criteria identified below

The data include the period between October 2000 and July 2011. This is the only time span for the data provided by the Central Bank of Brazil at this stage. The records were published monthly by the Central Bank of Brazil as required by Art. 66 in Decree nº 55.762 of 17 February 1965, modified by Decree nº 4.842 of 17 September 2003. The records were published on the bank’s website starting October 2000, as per communique nº 011489 of 7 October 2003. This remained the case until August 2011, after which the amount of each transaction was no longer disclosed (and publication of these stopped altogether after October 2011). The disclosure of the records was suspended in order to review their legal and technical aspects, and ensure their suitability to the requirements of the rules governing the confidentiality of the information (Law nº 12.527 of 18 November 2011 and Decree nº 7724 of May 2012) (pers. comm. Central Bank of Brazil, 2016. Name of contact available upon request to Authors).

The records track transfers of foreign capital made from abroad to companies domiciled in Brazil, with information on the foreign company (name and country) transferring the money, and on the company receiving the capital (name and federative unit). For the purpose of this study, we consider the four categories of foreign capital transactions which are published with their amount and currency in the Central Bank’s data, and which are all part of the “Register of financial transactions” (abbreviated RDE-ROF): loans, leasing, financed import and cash in advance (see below for a detailed description). Additional categories exist, such as foreign direct investment (RDE-IED) and External Investment in Portfolio (RDE-Portfólio), for which no amount is published and which are therefore not included.

We used the data posted online as PDFs on the bank’s website, and created a script to extract the data automatically from these four categories into the RCE_Unclean_full_dataset.csv file. This data set has not been double-checked manually and may contain errors. We used a similar script to extract rows from the "cancelled transactions" sections of the PDFs into the RCE_Unclean_cancelled_dataset.csv file. This is useful to identify transactions that have been registered to the Central Bank but later cancelled. This data set has not been double-checked manually and may contain errors.
From these raw data sets, we conducted the following selections and calculations in order to create the RCE_Clean_selection_dataset.csv file. This data set has been double-checked manually to secure that no errors have been made in the extraction process.

We selected all transactions whose recipient company name corresponds to one of these nine companies, or to one of their known subsidiaries in Brazil, according to the list of subsidiaries recorded in the Orbis database, maintained by Bureau Van Dijk. Transactions are included if the recipient company name matches one of the following:
- the current or former name of one of the nine companies in our sample (former names are identified using Orbis, Bloomberg’s company profiles or the company website);
- the name of a known subsidiary of one of the nine companies, if and only if we find evidence (in Orbis, Bloomberg’s company profiles or on the company website) that this subsidiary was owned at some point during the period 2000-2011, and that it operated in a sector related to the soy or beef industry (including fertilizers and trading activities).
For each transaction, we extracted the name of the company sending capital and when possible, attributed the transaction to the known ultimate owner.

The name of the countries of origin sometimes comes with typos or different denominations: we harmonized them.

A manual check of all the selected data unveiled that a few transactions (n=14), appear twice in the database while bearing the same unique identification number. According to the Central Bank of Brazil (pers. comm., November 2016), this is due to errors in their routine of data extraction. We therefore deleted duplicates in our database, keeping only the latest occurrence of each unique transaction. Six (6) transactions recorded with an amount of zero were also deleted. Two (2) transactions registered in August 2003 with incoherent currencies (Deutsche Mark and Dutch guilder, which were demonetised in early 2002) were also deleted.

To secure that the import of data from PDF to the database did not contain any systematic errors, for instance due to mistakes in coding, data were checked in two ways. First, because the script identifies the end of the row in the PDF using the amount of the transaction, which can sometimes fail if the amount is not entered correctly, we went through the extracted raw data (2798 rows) and cleaned all rows whose end had not been correctly identified by the script. Next, we manually double-checked the 486 largest transactions representing 90% of the total amount of capital inflows, as well as 140 randomly selected additional rows representing 5% of the total rows, compared the extracted data to the original PDFs, and found no mistakes.

Transfers recorded in the database have been made in different currencies, including US dollars, Euros, Japanese Yens, Brazilian Reais, and more. The conversion to US dollars of all amounts denominated in other currencies was done using the average monthly exchange rate as published by the International Monetary Fund (International Financial Statistics: Exchange rates, national currency per US dollar, period average). Due to the limited time period, we have not corrected for inflation but aggregated nominal amounts in USD over the period 2000-2011.

The categories loans, cash in advance (anticipated payment for exports), financed import, and leasing/rental, are those used by the Central Bank of Brazil in their published data. They are denominated respectively:
“Loans” (“emprestimos” in original source) - : includes all loans, either contracted directly with creditors or indirectly through the issuance of securities, brokered by foreign agents.
“Anticipated payment for exports” (“pagamento/renovacao pagamento antecipado de exportacao” in original source): defined as a type of loan (used in trade finance)
“Financed import” (“importacao financiada” in original source): comprises all import financing transactions either direct (contracted by the importer with a foreign bank or with a foreign supplier), or indirect (contracted by Brazilian banks with foreign banks on behalf of Brazilian importers). They must be declared to the Central Bank if their term of payment is superior to 360 days.
“Leasing/rental” (“arrendamento mercantil, leasing e aluguel” in original source) : concerns all types of external leasing operations consented by a Brazilian entity to a foreign one. They must be declared if the term of payment is superior to 360 days.
More information about the different categories can be found through the Central Bank online.

(Research Data Support provided by Springer Nature)


The authors would like to acknowledge the support from the Erling-Persson Family Foundation through the Global Economic Dynamics and the Biosphere Programme (GEDB) at the Royal Swedish Academy of Sciences, the Futura Foundation, Vinnova (Earth System Finance), Mistra Financial Systems (MFS), the Nippon Foundation (Nereus Program – Predicting the Future Oceans) and the Foundation for Strategic Environmental Research (Mistra) through the Stockholm Resilience Centre (Stockholm University).


Associated Publication

Galaz, V., Crona, B., Dauriach, A., Jouffray, J.-B., Österblom, H., Fichtner, J. 2018. Tax havens and global environmental degradation. Nature Ecology & Evolution (in press). doi:10.1038/s41559-018-0497-3

Reference to DMP - Data Management Plan