Board social capital and structure, ownership and financial variables of Brazilian companies: A three levels dataset integrating directors, board networks and firm characteristics

This data article incorporates, in an unbalanced panel data, five variables types: financial and market; board structure; board network and social capital; ownership and governance level; the cost of capital. The dataset is formed of 6024 firm-level annual observations based on 622 Brazilian public companies investigated between the years of 2002 and 2015, totaling 56 variables. A three-level data structure was created to allow aggregate directors and network board data into the panel data. Directors' data and adjacency matrix are included to allow for multilevel hierarchical analyzes as well as the use of analytical methods of social networks.


Data
The unbalanced panel is formed of 6024 firm-level annual observations based on 622 Brazilian public companies investigated between the years of 2002 and 2015. Fig. 1 shows the distribution of the number of companies and the market capitalization annually. We identified all the companies listed on B3 (former BM&F Bovespa), the Brazilian stock exchange, which effectively operated during the analyzed period, regardless of market liquidity.
Data on the Brazilian capital market are very fragmented and not easily collectable. For this reason, the dataset was constructed using five different sources: 1) B3, the only Brazilian stock exchange; 2) Brazilian Securities Commission (CVM); 3) Economatica® databank, an application that brings together the largest amount of market and financial information of Brazilian companies; 4) Thomson Reuters Eikon databank; 5) JP Morgan website www.adr.com. Even using multiple sources, the most relevant variables in the dataset had to be manually collected, filtered and analyzed before they were useful for firm-level analysis. These and all other variables are detailed in the next section.
In the panel data, we have identified 56 variables and attributes of the companies that are available in the supplemental material labeled "1_Panel_Data_Brazilian_Companies". Data is available in STATA, SPSS and Excel files. To facilitate the understanding of how the variables were operationalized, we divided the description into six blocks: company identification (Table 1); financial and market variables ( Table 2); board structure variables (Table 3); board social capital and network variables (Table 4); ownership and governance level variables ( Table 5); cost of capital variables (Table 6). In the dataset, the variables are listed in the same order as the tables, with the same name. It should be stressed that not all variables present the same number of cases since many of them depend on market liquidity and analysts' evaluation to be produced. Therefore, in Tables 2e6 we list the number of valid cases.
Specifications Table   Subject area Business, Management, and Accounting More specific subject area Business, Management, and Accounting (General) Type of data Panel Data Table  How data was acquired We collected raw data from five sources: B3 Brazilian stock exchange, Brazilian Securities Commission (CVM), Economatica, Thomson Reuters Eikon, and JP Morgan.

Data format
Raw, filtered and analyzed.

Experimental factors
A sample of companies listed on the Brazilian stock exchange.

Experimental features
The panel data incorporates at the firm level five variables types: financial and market; board structure; board network and social capital; ownership and governance level; the cost of capital. Data at the director and network levels are embedded in the data article.   Indicates the year in which the variables are measured, acting as the basis of time measurement.
Source: Economatica®. Real-valued ratio Return on Asset (ROA): The ratio between the company's profitability and the total volume of its assets. It is the company's ability to use its assets to generate profit. The data was consolidated annually (2002e2015), with reference month set in December of each year. The ROA is also known as LAJIRDA, which means earnings before interest, income tax, depreciation and amortization on total assets. This measure is obtained through the equation: 5976 Ln (ROA) ¼ Ln (LAJIR/AT), wherein: LAJIR ¼ profit before interest and taxes; AT ¼ book value of total assets. Market_Value Real-valued numeric Market value is the amount that stock market investors are willing to pay for trading on stock exchanges related to a specific company. It is obtained by multiplying the unit value of shares by the total number of shares that make up capital stock. The amounts are presented in thousands of reais (R$), adjusted by the inflation indexes.

4228
Beta Real-valued numeric The Beta Index is an indicator that measures the sensitivity of an asset to the behavior of a portfolio that represents the market. It is a measure of the risk that an investor is exposed to when investing in a particular asset compared to the whole stock exchange market.

3375
Beta ¼ (Covariance between Return on Asset and Market)/ (Variance of Return on Market) Beta High: Beta> 1 Beta Neutral: Beta ¼ 1 Beta Low: Beta <1 Current_Liabilities Real-valued numeric Current liabilities are obligations normally paid within one year: accounts payable, debts with suppliers of goods or raw materials, taxes to be collected (for the government), bank loans due in the next 360 days, provisions (expenses incurred, generated, not yet paid but already recognized by the company: income tax, vacation, 13th salary, etc.). Values consolidated annually and adjusted for inflation. Data in thousands of reais (R$).

6017
Long_Term_Liabilities Real-valued numeric Long-term liabilities are debts of a company which will be settled after the end of the following financial year. In most

Experimental design, materials, and methods
Due to the large number of variables, whose experimental design is different for each one of them, we specify in the description column of Tables 1e6 how they were operationalized. Board structure variables (Table 3) and social capital board and network variables (Table 4) have a more complex design. For this reason, we use two auxiliary datasets. In Fig. 2, we represent how each of the datasets relate to composing the panel data variables.  [3] indicator measure the performance of a particular company based on the sum of the market value of its shares, plus its debts, which is divided by the book value of its total assets.

4622
Consolidated annual sales amount (in thousands of reais). In order to define the sales growth of 2002, it was necessary to collect data for 2001, even though it was outside the period determined for this study. The variable was operated using the formula: Sales Growth (t-1) ¼ Sales Volume (t-1) -Sales Volume (t))/ (Sales Volume (t-1)) Source: Economatica®.  To build the board structure variables, we need data about the directors in each of the years. Director-level data is available in the "3_Directors_Data" dataset, which points out the directors at each company, as well as some attributes that were essential in defining the board structure (total of 67,957 records).
We generated the board network and social capital variables also from the directors' level data. First, we conceived an affiliation matrix in a 2-mode format for each year studied, crossing companies in one mode with directors in another. Then, through the UCINET® software, this 2-mode matrix was converted into a 1-mode matrix at the company level. In such an adjacency matrix, two companies are directly linked if they shared at least one director (board interlocking). Therefore, the presence of ties was defined in the binary form in the matrix cells. We then used UCINET® software to generate relational variables at the company level, later aggregated into panel data (see Table 4 for details). The adjacency matrices valued for each year are in the supplementary material: "2.1_Board_-Social_Capital_Network_Data", whose value greater than 0 represents the market value of a company alter (column) in relation to a particular company (ego); the adjacency matrix valued "2.2_Board_-Heterogeneous_Social_Capital_Network_Data" considers the same market value of the company alter,

Continuous
Resources immersed in the network of social relations between Boards of Directors through the so-called board's interlocks, accessed or/and mobilized in intentional actions through the interaction of the network and the resources present in it [4]. Social capital is constituted by the set of relations that give an individual access to resources that do not necessarily belong to him and that he would not have access were it not for their relationships. Access is carried out through so-called boards interlocks, in which a network rich in social capital is a network rich in mobilizable resources. The operationalization of Social Capital is given by the sum of the relational resources present in direct relationships. The ties of the ego network were identified, that is, the direct ties of each company with the others through the board interlocks. Soon after, the market value of each of the company's relationships (total value of the shares traded on the stock exchange) was identified. Finally, the value of the relational resources of each tie of the company was added. The financial data was collected on Economatica software. Values in thousands of Reais (R$).

Continuous
To reach indirect relations, we performed the structural holes procedure, which generated the redundancy value of each direct relation in relation to each of the network companies. The concept of redundancy is based on the following principle: If A is connected to B and C, and B is connected to C, the tie from A to B is redundant, because actor A can influence B through C. The measure of redundancy calculates, for each actor, how much of the other actors in the network are also connected to another actor. To say that the tie from A to B is highly redundant means that most of the other actors in the network also have a tie with B. Actors in networks with high redundancies are actors that are in networks with few "structural holes." We then reduce the value 1 of the redundancy value found for each company (of each Alter), thus generating a heterogeneity score for each alters. The greater the heterogeneity, the greater the number of structural holes present in the network. As the last step, this heterogeneity value was multiplied by the market value of each existing tie. Finally, the value of the relational resources of each tie of the company was added. The financial data was collected on Economatica software. Values in thousands of Reais (R$).

6016
Source: Brazilian Securities and Exchange Commission (CVM), Reference Forms and Annual Information (IAN). Note: Network variables generated from adjacency matrices "2.1_Board_Social_Capital_Network_Data" and "2.2_Board_-Heterogeneous_Social_Capital_Network_Data", available at supplementary material. Variables processed by the UCINET® software. The concentration of property is measured by means of an adaptation of the Herfindahl-Hirschman Index (HHI). This index is usually used to measure the degree of competition in a particular industry, but it is also used as a measure of the concentration of ownership in a given company. Their values range from 0 to 1, where the higher the index, the higher the concentration. It is calculated by summing the square of the individual voting shares owned by the three largest shareholders. It is defined as HHI: On what: The concentration of property is measured by means of an adaptation of the Herfindahl-Hirschman Index (HHI). This index is usually used to measure the degree of competition in a particular industry, but it is also used as a measure of the concentration of ownership in a given company. Their values range from 0 to 1, where the higher the index, the higher the concentration. It is calculated by summing the square of the largest individual voting shares owned by the five largest shareholders. It is defined as HHI: , is the ownership percentage of owner i.

HHI_3_Largers_PF 4 Ratio
The concentration of property is measured by means of an adaptation of the Herfindahl-Hirschman Index (HHI). This index is usually used to measure the degree of competition in a particular industry, but it is also used as a measure of the concentration of ownership in a given company. Their values range from 0 to 1, where the higher the index, the higher the concentration. It is calculated by summing the square of the individual preferred shares owned by the three largest shareholders. It is defined as HHI: Ratio The concentration of property is measured by means of an adaptation of the Herfindahl-Hirschman Index (HHI). This index is usually used to measure the degree of competition in a particular industry, but it is also used as a measure of the concentration of ownership in a given company. Their values range from 0 to 1, where the higher the index, the higher the concentration. It is calculated by summing the square of the individual preferred shares owned by the five largest shareholders. It is defined as HHI:   The Ex-Ante R PEF cost of capital data was collected from the Thomson Reuters Eikon platform and found in the database in decimal values. According to Espinosa and Trombetta [7], the cost of capital R PEF is defined by the equation: On what: Weighted average cost of equity and third-party capital. In terms of the coefficient, the cost of capital represents a minimum rate that the company must obtain in its operations, which indicates the minimum necessary remuneration to be earned to maintain the value of its shares and the respective sustainable growth of the company. This way of measuring the cost of capital is called Ex-Post and assumes the following expression [8]: On what: WACC ¼ Weighted Average Cost of Capital (WACC); Tc ¼ rate of income tax and social contribution of the legal entity; E ¼ market value of equity (company) or shareholders' equity (in R$); D ¼ market value of the company's third-party capital (in R$); V ¼ E þ D (market value of total capital, in R$); E/V ¼ ratio of equity to the total financing of the company (in market values); D/V ¼ proportion of third-party capital over the total financing of the company; Re ¼ Equity Rate -Capital Assets Pricing Model or Capital Asset Pricing Model (CAPM). CAPM ¼ Risk free rate þ [Beta * country risk premium]; Rd ¼ cost rate of third-party capital before income tax. The cost of thirdparty capital is defined in accordance with the onerous liabilities identified in the loans and financing maintained by the company. It is calculated using consolidated government rates, debt adjustment factor and the ratio of long-and short-term debt to total debt. Rd ¼ ((Short-term debt) * Short-term debt pre-charge þ Long-term debt * Long-term debt pre-charge)/Total debt) * (1-Tc).

2628
Source: Thomson Reuters Eikon. Note: Cost of capital variables produced by authors.