The idea of disclosure addresses the publishing of qualitative or quantitative accounting information, either compulsory or voluntary, that provides accurate and useful information for decision-makers (Lopes & Alencar, 2010). Similarly, Frenkel, Guttman, and Kremer (2020), and Acar and Okzan (2017) propose that disclosure is an important mechanism for accounting information quality, which has gained importance in recent years within the accounting research field.

A detailed discussion on the topic was proposed by Verrecchia (2001), who presents three (or four) broad categories of study. The first is composed of association-based disclosure studies, which address the association between investors' behavior and disclosure. The second category consists of discretionary-based disclosure, which seeks to establish the reasons for disclosure by investigating the reasons that lead companies or managers to disclose certain information.

The third category encompasses efficiency-based disclosure studies, which seek to investigate which information is preferred and most efficient, before disclosure. Finally, there is also the relationship between information retention and managers' motivation for disclosure. In this regard, the market assimilates the omitted information's content and, when it does not match managers' intention, it is defined as the disclosure level threshold (equilibrium point).

In this study, we consider discretionary-based disclosure following the classification of Verrecchia (2001), or voluntary disclosure theory, as established by several authors, which assumes that managers have incentives to voluntarily disclose (or not disclose) information (Klann & Beuren, 2011; Rufino & Monte, 2015). Disclosures should happen voluntarily if the benefits generated by such disclosure are higher than the underlying costs (Murcia & Santos, 2009).

However, not all companies will have high levels of voluntary disclosure. Haddad et al. (2020) highlight that the voluntary disclosure of accounting indices provided in the manufacturing companies' annual reports in the Gulf Cooperation Council was considered extremely low. A possible explanation for that may lie in the contractual theory of the firm. Under this perspective, accounting information is responsible for monitoring contracts and, consequently, mitigating information asymmetry and transaction costs. On the other hand, although the increase in voluntary disclosure improves agent monitoring, it tends to increase agent remuneration costs (Hermalin & Weisbach, 2012). Thus, the increase in disclosure generates an increase in transaction costs that, in the end, can overcome the benefits of disclosure and destroy firms' value (Hermalin & Weisbach, 2012).

Researchers have sought to identify variables related to companies' voluntary disclosure. Tsang, Xie, and Xin (2019), for example, reported that foreign investments lead to increased voluntary disclosure. Additionally, they found that foreign investments have a greater impact on voluntary disclosure than national investments. Guttman and Meng (2020) showed that companies' ability to disclose or omit voluntary information could affect their investment decisions. While Cassar, Gerakos, Green, Hand, and Neal (2018) point out that investors' agency costs influence voluntary disclosure.

Among these discussions, research has widely argued that companies' profitability level can influence voluntary disclosure by managers or companies. For Salotti and Yamamoto (2008), this variable fits the concept of adverse selection, since companies with better financial performance have greater incentives to disclose such performance. Otherwise, non-disclosure could lead the market to infer that the performance was unsatisfactory. Consequently, investors may not select a good investment option due to the non-disclosure of information.

In this sense, Alsharari and Alhmoud (2019) and Rouf and Akhtaruddin (2018) argue that there is a positive association between profitability and voluntary disclosure. That is, companies with higher profitability tend to disclose more information to increase investors' information level. Kolsi (2017) and Rufino and Monte (2015) corroborate this argument by stating that companies with higher profitability have greater incentives to disclose information about their financial situation. Therefore, high profitability should be understood by managers as good information. Consequently, managers should highlight with greater emphasis when companies' have greater profitability.

Similarly, Aly et al. (2018) highlight that several companies benefit from the relationship between profitability and voluntary disclosure to signal good economic performance to the market. In this scenario, Félix Júnior, Oliveira, Miranda, Lagiola, and Galvão (2018) and Haniffa and Cooke (2005) also address voluntary disclosure of profitability by relating it to the disclosure of voluntary information. For the authors, companies with higher profitability tend to disclose voluntary information to make their performance evident to the market.

These understandings are in line with the Signaling Theory, which, in a market context, explains voluntary disclosure as managers' attempt to mitigate information asymmetry by signaling the qualities that differentiate their companies from others to the market (Klann & Beuren, 2011; Wardhani, 2019).

In the international context, Acar and Ozkan (2017) tested the hypothesis that there is a difference in companies' corporate governance structure that recognize provisions. The study analyzed 191 non-financial companies listed on BIST (Borsa Istanbul), the Turkish stock exchange. The results showed that the companies that recognize provisions have a larger board of directors, and 62% of the observations recognized provisions. Among these companies, from 2005 to 2010, only 32% provided IAS 37 information with full disclosure requirements (post-IFRS).

In this same scenario, Acar and Ozkan (2017) state that the corporate governance structure impacts financial reporting, given the importance of corporate governance in reducing information asymmetry and agency conflicts. Consequently, corporate governance positively influences financial information disclosure and the quality of this disclosure. Allaya, Derouiche and Muessig (2018) highlight that companies with greater voluntary disclosure have more long-term debts, suggesting that companies benefit from an extensive disclosure through greater access to long-term debt.

Regarding the relationship between profitability and companies’ voluntary disclosure level, Bhatia and Dhamija (2015) investigated the relationship between the voluntary disclosure of financial indicators with the performance of companies listed on the Indian stock exchange. Among the indexes studied, profitability showed a negative relationship with disclosure, however, it was not significant. Nevertheless, Kolsi (2017) showed that profitability positively affects companies' voluntary disclosure in Arabia, which can help investors in decision-making and lead to a fair resource allocation.

Additionally, Alsharari and Alhmoud (2019) and Ahmadi and Bouri (2017) highlight the influence of profitability on voluntary disclosure. Within this scenario, through the positive association between profitability and voluntary disclosure, companies seek to make more clarifications to leverage the value that the disclosure of their voluntary information can generate (Félix Júnior et al., 2018). Therefore, a higher rate of voluntary disclosure by companies with higher profitability can improve their organizational performance level.

In Brazil, some studies that investigated the factors that influence voluntary disclosure found a positive, however, not significant association between profitability and voluntary disclosure, such as the studies of Murcia and Santos (2009), and Klann and Beuren (2011). On the other hand, Consoni and Colauto (2016) and Angonese et al. (2014) found a negative association, however, also not significant.

Regarding the relationship between performance-related measures and voluntary disclosure, Haddad et al. (2020), Karajeh, Ibrahim and Lode (2017) and Egbunike and Tarilaye (2017) showed that a higher level of financial leverage and firm size are associated with more voluntary disclosure of accounting ratios. Furthermore, Elfeky (2017) tested a theoretical framework showing a positive significant correlation between firm size, firm profitability, firm leverage, independent directors and auditor type, and the extent of voluntary disclosure of Egyptian companies.

The studies of Rouf and Akhtaruddin (2018), Aly et al. (2018), Elfeky (2017), Rufino and Monte (2015), and Haniffa and Cooke (2005) stand out among the articles that found a positive correlation with statistical significance on the relationship between profitability and the level of voluntary disclosure in literature. However, this relation, that is, profitability versus voluntary disclosure, still needs to be understood in the Brazilian scenario, where research has shown divergent results and no statistical significance, opening research avenues for further studies of this nature. Thus, we propose the following research hypothesis:

H1: higher profitability is associated with a higher level of voluntary disclosure of companies.

This study's non-probabilistic sample comprises all B3 non-financial companies, which encompass 527 companies within the 2010-2018 period (9 years), resulting in a total of 4743 observations. We opted to collect data from 2010, since it was the first year of publication of the Reference Form (FR) and 2018, as it was the latest year of available information.

The Reference Form (FR) was instituted by the Securities and Exchange Commission (CVM) through the CVM Instruction NO. 480/2009 and is disclosed annually by companies that issue securities. This instruction was defined as the data source of voluntary disclosure due to its uniform structure, contributing to the comparability of the information disclosed. This document has also set a new standard of mandatory disclosure requirements, which may differ from voluntary disclosure results of previous research (Consoni & Colauto, 2016).

Despite the RF being mandatory, the data considered for creating the voluntary disclosure variable was the information that allowed issuer identification. This procedure is in line with recent national research on the topic (Consoni & Colauto, 2016; Consoni, Colauto & Lima, 2017).

The observable variables (Metrics) were collected with the aid of Economatica Software. Companies from the financial sector were excluded from the sample because they could bias the results, given their specific characteristics not comparable with non-financial companies.

Our dependent variable Voluntary Disclosure was collected through content analysis of FR reported on the Brazilian Securities and Exchange Commission website for B3-listed companies for the years from 2010 to 2018. According to Bardin (2010), the content analysis technique consists of procedures of analysis and description of the database researched, which allows the inference of information about the sample studied in the research.

To measure the Voluntary Disclosure proxy, we built a Voluntary Disclosure Index based on the definition of categories and subcategories of the elements that compose the Voluntary Disclosure of companies proposed in the study of Murcia and Santos (2009). The following subcategories were excluded from the original proxy of Murcia and Santos (2009): Statement of Cash Flow, Segment reporting, Foreign Exchange Exposure, and Director's Remuneration. We opted to exclude these elements because the Brazilian Accounting Standards Committee (CPC) made their disclosure mandatory. Thus, such information is no longer considered of voluntary disclosure. The resulting proxy is composed of 26 items of Voluntary Disclosure, as shown in Table 1.

Table 1
Elements used in the voluntary disclosure proxy

Source: Adapted from Murcia and Santos (2009).

Based on the subcategories described in Table 1, the level of Voluntary Disclosure was calculated by the number of subcategories disclosed by each company, divided by the total number of subcategories (26). Therefore, if a company discloses all 26 subcategories, it will have a Voluntary Disclosure Index of 100%. If a company discloses 13 subcategories, its Voluntary Disclosure Index will be 50%, and so on. Equation 1 describes this approach.

(1)

Where:

DISCV = Firm’s voluntary disclosure;

SD = Disclosed subcategories (1 if the subcategory is disclosed, 0 if not); and

TS = Total Subcategories (26).

The independent variable of interest analyzed was Return on Assets (ROA). Thus, we expect that more profitable companies will disclose more information to legitimize their existence. Therefore, firms with higher ROA should present a higher level of voluntary disclosure (Elfeky, 2017; Haniffa & Cooke, 2005). The data collection source for this metric was the software Economatica. The variable was calculated through the ratio between the firm's Net Profit and Total Assets.

We also included in the econometric model the five most used control variables by the literature. The inclusion of control variables avoid sample bias and guarantees the robustness of the research's empirical analysis, as shown in Table 2 (section 3.4).

As shown in Equation 2, we estimated the following econometric model to address this research's objective, which consists of evaluating whether profitability positively influences firms' voluntary disclosure level.

(2)

The description of the dependent and independent variables used in the econometric model are shown in Table 2.

DISCV was obtained through Equation (1) according to section 3.2. The term "Ɛ" also present in the econometric model, represents unobserved factors that can affect the dependent variable. The Multiple Linear Regression data analysis method was applied with fixed effect panel data, as it is the most appropriate data procedure for continuous dependent variables (Wooldridge, 2002).

The following procedures were adopted for data preprocessing: a) missing values ​​were removed from the sample with the software Stata since some companies stopped disclosing certain information. Thus, companies that did not disclose any of the variables used in this article were eliminated from the sample; b) to remove outliers from the sample, avoiding potential bias to the results, first, data were rearranged in ascending order, and then 1% of the companies with the lowest and highest values ​​were excluded. After these procedures, our initial sample of 4,743 observations from all B3 non-financial companies from 2010 to 2018 was reduced to a final sample of 2,783 observations.

Table 2
Variables of the econometric model used in the article

Source: Elaborated by the authors.

Our study presents a methodological limitation as the measurement of the voluntary disclosure proxy can be a weakness given its subjectivity. This limitation arises from the method used to build this variable - content analysis – which requires a subjective analysis. However, despite such possible arbitrarity in constructing any non-measurable variable, several studies in this field have adopted the same criterion for this proxy.