Impact of information disclosure on the value of shares of mixed economy companies

Impacto da evidenciação de informações no valor das ações das sociedades de economia mista

The purpose of the Disclosure Theory is to explain phenomena arising from the disclosure of financial information (Salotti & Yamamoto, 2005). One of the perspectives studied is the impact on the value of company shares resulting from the disclosure of information. For some authors, such as Gibbins, Richardson and Waterhouse (1990) and Von Alberti-Alhtaybat, Hutaibat and Al-Htaybat (2012), corporate disclosure is a far-reaching term, that deals with the disclosure of information by physical or digital means that is either financial or otherwise, numerical, qualitative, mandatory or voluntary. The involuntary disclosure of information can also be included in this definition (Skillius & Wennberg, 1998). Verrecchia (2001), on the other hand, states that there is no unified or integrated disclosure theory, since there is no central paradigm underlying all research, which examines small parts of the whole.

Healy and Palepu (2001) argue that organizations can disclose mandatory information, for example, by way of financial reports, voluntary information by way of press releases and corporate reports and, involuntarily information via financial analysts and the press. They further claim that such disclosure is essential for the capital market to function efficiently.

Mixed-capital companies (SEM) in Brazil are relevant economic agents and present in significant numbers, since they represent one of the ways government acts in the economy (Ribeiro & Alves, 2006). However, in addition to the various problems of government inefficiency in sectors that are vital to the country’s development, mixed-capital companies face risks of political influence, which can affect the value of those companies in which the government is an investor (Kuschbrey et al., 2014), but the value of a company’s shares can also be affected by its own and by third party disclosures, as can be seen in research into information transfer.

This research, therefore, aims to analyze whether disclosure of the firm’s mandatory and voluntary information, and the disclosure of information about the government affects the share price of companies. Whether the disclosure of this information affects mixed-capital companies differently is also analyzed.

The justification for this research is the importance of obtaining a better understanding of the Brazilian stock market, the behavior of shares and the reaction of investors to the disclosure of information. It also examines the relevance and functionality of accounting, market communication and sovereign rating classifications as a way of reducing information asymmetry between companies and investors. Brazil is a favorable scenario for such an analysis because of the strong participation of the government as an investor in large companies that are quoted on the Brazilian stock market.

The data of 383 companies listed on the B3 [Brazilian Stock Exchange] between 2003 and 2015 were analyzed; more specifically the abnormal returns of firms were calculated that were observed close to events disclosing mandatory information, such as the publication of regular financial statements (DFPs) and quarterly information (ITRs); voluntary information, as communicated to the market, and the transfer of information, such as the disclosure of the sovereign rating. The results show that mixed-capital companies and other companies are affected equally with regard to mandatory disclosure. With regard to voluntary disclosure and sovereign rating, however, there is an additional effect on mixed-capital companies when compared to other companies.

This study contributes in at least two respects to the accounting and finance literature. First, it adds to the discussion about the effect of the information released by the company on its share price, thus providing the Brazilian stock market with empirical evidence. Secondly, it contributes to debates about information transfer as it relates to an important shareholder in the Brazilian stock market, the government. This research also contributes to practical aspects, as it demonstrates how the market behaves in the face of the disclosure of corporate information and the sovereign rating, enabling investors to have a greater understanding of disclosure events in the Brazilian scenario.

The field of study for this research was Brazilian companies that are listed on the B3. It was conducted to analyze the impact on the value of company shares resulting from the disclosure of company information. The sample also comprises the mixed-capital companies described in Table 1. For the statistical tests, Banco do Estado de Santa Catarina S.A. (Besc) was removed from the sample, due to the lack of observations for the necessary variables.

Table 1
Sample description

Source: research data.

The data used were taken from the times when the disclosure events occurred. They were used as milestones for defining the periods to be used for analyzing the value of the shares. Data for disclosure of the SFSs, QIs and market notices were collected manually from the CVM website, while information from the sovereign rating disclosure events, issued by Standard & Poor’s, Fitch Ratings and Moody’s, was collected from the website of the Brazilian Department of the Treasury.

The period analyzed in the survey is from the beginning of 2003 until the end of 2015; 2003 is the beginning of the period for which data related to communiques to the market are available for research on the CVM website. The variables analyzed are the cumulative abnormal stock returns in the period defined by the disclosure events. The event window, the period in which the returns for each event were collected, and which is subject to a degree of personal choice (Camargos & Barbosa, 2003) was defined as being one day before the event and one day after the event, in accordance with Beyer et al. (2010).

To calculate returns in the research, the form of discrete capitalization was used, in which the stock returns were obtained by equation (1).

where r is the rate of return, P_t the share price in period t and P_(t-1) is the share price in period t-1.

For abnormal returns, this research used the market-adjusted return model, in which abnormal returns are calculated by the difference between the observed return of the firm and the return of a market portfolio in the same period, according to equation (2). In Brazil, the most widely used market index is the IBOVESPA (Camargos & Barbosa, 2003), but in this study the average daily value-weighted return of all companies listed on the B3 was used, weighted by the company’s market value in relation to the total.

where A_it is the abnormal return of a given stock i in time period t, R_it is the observed return of the same share i in the same time period t and R_mt is the return of the market portfolio in period t.

Cumulative Abnormal Return (CAR) (3), which consists of accumulating the abnormal returns observed in the event window by means of their sums, was used to accumulate the abnormal returns on the days of the event window, since there is no way of specifying the moment when the market received information about the events (Camargos & Barbosa, 2003).

where CAR_i is the cumulative abnormal return of asset i, t₁ is the first day of the event window, t₂ is the last day of the event window and A_it is the abnormal return of a given stock i in time period t.

For the statistical test, a regression was performed, similar to the one used by Beyer et al. (2010):

where CAR_i,t is the cumulative abnormal return of asset i, in year t, car_i,t^DO is the three-day cumulative abnormal return centered on the mandatory disclosure announcements of the financial statements, car_i,t^DV is the three-day cumulative abnormal return centered on the voluntary disclosure announcements to the market and car_i,t^GOV is the three-day cumulative abnormal return centered on the government’s sovereign rating announcements. To control for potential effects associated with the size, profitability, and growth of companies, control variables were added to the model. The variable Vm is the Neperian logarithm of the market value of company i in period t. The variable Roa represents the return on assets of company i in period t. The Book-to-Market of company i in period t is represented by the variable Btm. IFRS is a dummy worth 1 in 2010 and later and 0 otherwise, to reflect the possible impacts of accounting changes.

In cases where there was no announcement, the variables car_i,t^DO , car_i,t^DV and car_i,t^GOV took the value 0 and in cases where more than one event occurred, the returns for each window were added.

To identify the additional impact of mixed-capital companies on disclosures, interactions between the variables and a dummy variable that identifies this class of company were added to the model.

where SEM is a dummy variable that assumes value 1 if the company is a mixed capital company and 0 otherwise. The other variables follow the same definition as in equation (4).

Table 2 shows the sample’s descriptive statistics demonstrating the relevance of mixed-capital companies by way of the mean difference test, since these companies have a significantly higher mean for Vm, Roa and Btm than other companies.

Table 2
Descriptive statistics

Source: research data. Note: Where CAR_i,t is the cumulative abnormal return of asset i, in year t, Vm is the market value of companies in Br. Reais, Roa is the return on the asset and Btm is the Book-to-Market..

Table 3 shows the correlations between the control variables of the models and the dependent variable, showing a significant correlation between the dependent variable and the Btm variable, between the Vm variable and the Roa and Btm variables, and between the Roa and Btm variable.

Table 3
Correlations of control variables

Source: research data. Note: Where CAR_i,t is the cumulative abnormal return of asset i, in year t, Vm is the market value of companies in Br. Reais, Roa is the return on the asset and Btm is the Book-to-Market.. * significant at 10%.

For the statistical tests, 3514 observations were used per company and per year, of which 239 observations come from mixed-capital companies. It is important to point out that there the researchers did not have that observations for all companies/year, and that observations are risk being ignored when control variables are added because of the data limitation of these variables.

The variables CAR_i,t , car_i,t^DO, car_i,t^DV, car_i,t^GOV, Vm, Roa and Btm underwent a Winsorization process with a lower limit of 1% and an upper limit of 99%, which consists of replacing the values below and above the limit determined by the lowest and highest remaining values, respectively. It should be noted that the results are similar without the Winsorization process.

Table 4 shows the regressions between the variables, in simple and interacting models, in addition to the controls. Model 1 relates the cumulative abnormal return for the year to the cumulative abnormal return around the mandatory disclosure, voluntary disclosure, and government rating events.

Model 2 adds the four control variables. In models 3 and 4, in addition to the variables listed in models 1 and 2, there are the interactions of these variables with the SEM variable, representing the additional value of the variables when they are related to SEM (Table 4). All models were estimated using a robust standard error for controlling heteroscedasticity.

Table 4
Result of regressions

Source: research data. Note: CAR_i,t is the cumulative abnormal return of asset i, in year t, car_i,t^DO is the three-day cumulative abnormal return centered on mandatory disclosure announcements of the financial statements, car_i,t^DV is the three-day cumulative abnormal return centered on voluntary disclosure announcements to the market and car_i,t^GOV is the three-day cumulative abnormal return centered on the announcement of the government’s sovereign rating, SEM is a dummy variable that assumes value 1 if the company is a mixed capital company and 0 otherwise, Vm is the Neperian logarithm of the market value of the company, Roa represents the return on assets of the company, Btm is the Book-to-Market and IFRS is a dummy worth 1 in 2010 and later and 0 otherwise. In cases where there was no announcement, the variables car_i,t^DO , car_i,t^DV and car_i,t^GOV took the value 0 and in cases where more than one event occurred, the returns for each window were added. *** significant at 1%.

Analyzing model 1 in Table 4, both the mandatory disclosure variable and the voluntary disclosure and government rating classification variables are significant at the 1% level. This result shows that these variables have an impact on the dependent variable, which is the annual cumulative abnormal return. It appears that the Vm and Btm variables were also significant. These results are in line with hypotheses 1, 2 and 3a of this research.

In model 3, as with models 1 and 2, the mandatory disclosure, voluntary disclosure and government rating classification variables are significant at the 1% level. There is no significance in the interaction variables for the interaction of the SEM variable with the mandatory disclosure variable, but there is significance in the interaction with the voluntary disclosure and sovereign rating classification variables. This result demonstrates that there is an additional impact for SEM in voluntary disclosure and sovereign rating events.

This fact is evidence of hypothesis 3b of this study, that is, that mixed-capital companies are affected differently than other companies when the sovereign rating is disclosed. In models 2 and 4, the control variables Vm, Btm and IFRS were also significant.

To establish robustness, tests were performed only with companies in the Finance and Insurance sector, the sector that contains the most SEM; another test was conducted excluding this sector. All the methodological details used in the previous estimates were maintained. The results were similar for both the Finance and Insurance sector and when this sector was excluded.

In short, in all models the cumulative abnormal returns variables around mandatory and voluntary disclosure and government classification rating events have a significant impact on annual cumulative abnormal returns. In addition, in all models with interactions, the interactions of the SEM variable with the voluntary disclosure and sovereign rating variables showed significant results, while the variables Vm, Btm and IFRS were significant in the models with controls.

These results corroborate the research of Malaquias and Lemes (2015) who found there was an impact for mandatory disclosure, but they are contrary to the research by Mugaloglu and Erdag (2011), Sadeghzadeh and Karimi (2010) and Marques et al. (2010) who were unable to detect any impact. As for voluntary disclosure, our results are supported by the research conducted by Murcia et al. (2011), Souza et al. (2014) and Lima et al. (2012). The results also confirm research by Murcia, Murcia and Borba (2013), Creighton, Gower and Richards (2007) and Jorion, Liu and Shi (2005) that found a relationship between share value and ratings.

This research aimed to verify whether the disclosure of information has an impact on the value of company shares. It specifically tried to establish whether the disclosure of mandatory corporate information through the publication of financial statements, and voluntary information by way of notices to the market, and government information, by way of the sovereign rating classification affects the value of the shares of public companies and has an additional impact on mixed-capital companies.

To do so a study similar to that of Beyer et al. (2010) was undertaken to verify whether the disclosure of information has an impact on the value of company shares. In view of the results of the statistical tests, all the hypotheses raised in this study were confirmed. The conclusion drawn, therefore, is that investors react to the disclosure of financial statements, and market and government rating announcements in the Brazilian market, and to voluntary disclosure and government rating announcements, albeit in a different way when it comes to mixed-capital companies.

The research contributes towards a better understanding of the Brazilian capital market and of the behavior of investors, showing that the means of disseminating information are relevant and act to reduce information asymmetry. The study also contributes to the literature on disclosure and information transfer, corroborating some of the previous studies on this matter and developing the relationship between information disclosure and financial variables. It also indicates new research possibilities.

This study has its limitations, such as: its field of study, which covers only companies listed on the B3; the types of information that were used as events, which were just the disclosure of financial statements, market notices and the sovereign rating; the way in which abnormal returns are measured, which, of the many models available, uses the market adjusted model; and the subjective nature of the choice of the event windows.

Suggestions for future research include using other fields of study, for example, segmenting companies by sector, and adopting other types of information, such as relevant facts and media news. Event studies also enable different methods of measuring abnormal returns to be used, so this research can be conducted employing other methods and other event windows.