Accessing financial reports and corporate events with GetDFPData/Acessando relatorios financeiros e eventos corporativos com GetDFPData.
| Autor | Perlin, Marcelo S. |
-
Introduction
Financial statements portray the current financial situation of a company and are at the heart of corporate finance studies and investors' allocation decisions. Easy access to information is a mandatory quality of a well-functioning market. In Brazil, financial reports of companies traded on B3 (formerly known as BM&FBovespa), are available on its website (1) or on the CVM (Comissao de Valores Mobiliarios) website (2). The documents are organized into four systems: ITR/DFP for quarterly/annual financial reports, FRE for corporate events and FCA for a company's registry information. On the B3 website, one can find a simple web interface for accessing this large dataset for a single company.
However, gathering and organizing data for large-scale research, with many companies and many years, requires a significant amount of manual and tedious work. Financial reports and corporate events such as the balance sheet, dividend payments, and others must be downloaded or copied individually for each time period and later aggregated. Changes in the format over time can make this process slow, unreliable, and unreproducible.
Many commercial data vendors offer solutions for accessing clean financial statements and corporate events. Our bibliometric research demonstrates that, in a group of recent publications, approximately 70% of these studies used data from Economatica (3). Using data from commercial vendors requires the purchase of a license and may not be within reach of some research institutions and retail investors. Moreover, most data vendors do not offer a data format suitable for large-scale research or an API for easier access. Likewise, even when the data is available, it still requires reorganization, adding an unnecessary cost. These types of barriers to data acquisition may impede the development of local research in corporate finance and accounting.
GetDFPData provides an open and free interface to all financial statements distributed by B3 and CVM (Comissao de Valores Mobiliarios). It not only downloads the data but also cleans it, adjusts for inflation, and prepares it for research within a tabular format. Users only need to select companies and a time period to down-load all of the available data. Up-to-date information about companies traded on B3, including situation (active or inactive), sector, CVM code, and related tickers of traded assets, are available to the user. This information allows researchers to maintain a continuous workflow, from the selection of companies to hypothesis testing, facilitating and increasing the reproducibility of studies.
The software was created by researchers and designed for researchers. The dataset resulting from the application was formatted in line with modern methodologies for data organization and processing, the so-called tidy data (Wickham, 2014). Its main principle is to keep data in a row-oriented tabular format. This facilitates visualization and modeling of large financial datasets, without the need for further reorganization.
As with any provider of financial data, quality is key. However, having said that, it is important to understand that all data imported with GetDFPData has been taken directly from the source, B3's DFP and FRE system. Moreover, as is well-known, companies must report and be responsible for the information they provide. Financial markets and institution set strong and direct incentives for companies to release truthful information.
The package works as a mirror for all financial documents available in B3. Unlike other data vendors, the information is downloaded and read directly from the exchange. There is no intermediating compiled database. Therefore, the package benefits from the strong incentives that companies have to provide quality data. However, it is best to be aware that, in the current version, data from GetDFPData is offered as-is, without any error checking.
The main contribution of the software is to make it easy to access financial statements in large-scale studies by setting an open standard of data acquisition and organization (Gandrud, 2013). Researchers and investors unfamiliar with R can use the web version (4) of GetDFPData to download an Excel spreadsheet or CSV files with all the data. The software not only contributes to easier, unrestricted access but also provides more information than most commercial data vendors. History of corporate events such as dividend payments, stock splits and reverse splits, changes in stockholder composition, auditing reports, a company's remuneration policy, family relationships in the company, debt composition, history of governance listings and more are included in the output. The available content should satisfy a large variety of corporate finance and accounting studies.
This paper is organized as follows. First, we discuss the use of financial reports and software in a list of recent publications. There follow instructions for the web interface and a tutorial for installation and usage of GetDFPData in R. We next illustrate the use of our package with a reproducible example of a corporate finance study. We conclude the paper in the last section.
-
Literature Review
We surveyed the most important Brazilian journals to show the relevance of financial reports and corporate events data for research in the fields of accounting and finance: Contabilidade Vista & Revista, Revista Contabilidade & Financas, Revista de Contabilidade e Organizacoes, Revista Contemporanea de Contabilidade, and Revista Brasileira de Financas.
The list was built based on a simple rule: we select accounting and finance journals classified as "A2" in the "Qualis-CAPES" classification system for the period between 2013 and 2016, in the area of Business (Administracao, Contabilidade e Turismo). We believe that the accounting and finance fields of research would benefit the most from the data provided by GetDFPData. Therefore, we only consider journals which are strictly related to the Finance/Accounting fields, excluding others with more broad topics such as management. The "A2" classification is the highest tier for local journals. The only exception is Revista Brasileira de Financas, classified as "B1" (one rank below "A2"), which is included because it is the official publication of the Brazilian Society of Finance (SBFin) and because of its relevance as an outlet for financial research (Perlin & Portela Santos, 2015). To keep things manageable, we also restrict our survey to five years: 2013-2017.
In our survey, we look for papers that use financial reports of Brazilian firms as the main input for the data analysis procedure. We collect the following information for each paper that fits this requirement: authors, year of publication, journal title, subject, sample period, number of firms, number of observations, and databases used (data sources).
Table 1 presents the number of papers selected (surveyed) and the total number of papers published in each journal. Research based on financial reports and corporate events data of Brazilian listed companies is quite common. Of the published papers surveyed (477), 105 (22.01%) share this characteristic., Revista Contabilidade & Financas is the most important outlet for this type of research, both in absolute and relative terms. This journal published 34 papers based on financial reports and corporate events data of Brazilian listed companies and this number represents 28.81% of the papers published in the journal from 2013 to 2017. At the other extreme is Revista Brasileira de Financas, with 9 papers out of 85 (10.59%), showing a heterogeneous content in Finance.
Data sources used in the surveyed papers are listed in Table 2. As we can see, Economatica is the most used software: 77 out of 105 papers (73.33%) built their data sets from this commercial source. Far behind in this list are public data sources: CVM (23 papers, 21.90%) and B3 (5) (19 papers, 18.10%). The preference for Economatica can be likely explained by its data exporting tools. Although not in a suitable format for large scale research, Economatica allows the user to export data from multiple firms and several time periods at once. Whereas, on the CVM and B3 websites, data can only be retrieved for one company/period at a time.
In Table 3 we present descriptive statistics of three variables that characterize the samples used in the surveyed papers: number of observations, number of firms, and number of years. On average (at the median) the sample consists of 2,039 (722) observations, 159 (116) firms, and 7 (5) years. These variables vary widely, as indicated by the standard deviation, minimum, and maximum statistics. The characteristics of the samples indicate that researchers can greatly benefit from the proposed software, since it delivers data from several firms and time periods in formats ready to be analyzed by data analysis tools, including R, Stata, SPSS, and others.
Finally, in Table 4 we list the main subjects explored in the surveyed papers. Earnings management is the most common subject: 19 out of 105 papers (18.10%) deal with this research topic. Disclosure (14 papers, 13.33%), Performance (12 papers, 11.43%), IFRS (11 papers, 10.48%), and Corporate Governance (10 papers, 9.52%) are also very common subjects. Beyond the interest of the researcher and the importance of the subject, the choice of research topic is likely to be influenced by data availability. In this sense, our package could contribute to a greater diversity of research topics, since it provides access to data unavailable in other data sources for a large number of firms, such as executive compensation, family members in the company, transactions with related parties, auditing information, debt structure, and board composition. As shown in Table 4, papers dealing with these data are relatively rare in the Brazilian accounting and finance literature.
-
Using GetDFPData
The software is...
Para continuar a ler
PEÇA SUA AVALIAÇÃOCOPYRIGHT GALE, Cengage Learning. All rights reserved.
