3.1. Research Design and Variable Definition
The econometric model applied will be analyzed using the unbalanced panel data regression technique for fixed effects. This method allows controlling for unobserved heterogeneity in the cross-sectional units and estimating certain dynamic relationships (Wooldridge, 2010). Furthermore, it has the advantage of using a larger database than would be possible if only the same firms were repeated for all years. Thus, the results may present greater robustness and reliability. Based on the literature studied, a model was developed to test the study hypothesis. This model has as its dependent variable the AIV, which represents the indication of opportunistic insider trading by the firm. Thus, the econometric model that will be estimated is presented, according to Equation (1):
Equation (1) is presented with the aim of verifying whether aspects of corporate governance mitigate the signs of opportunistic insider trading, as described in the research hypothesis (H1). To increase the reliability of the results, a robustness test was performed. In this test, the variable AIV (Abnormal Idiosyncratic Volatility), which represents the sign of opportunistic insider trading, was replaced by the variable REPM (Tainted Reputation), which is represented by the number of republications of a firm. This variable was chosen because, according to previous studies, republication has already been used as a proxy to measure opportunism, given that firms with opportunistic insiders have a higher incidence of republications (Ali & Hirshleifer, 2017; Javakhadze, Pennathur & Silverstein, 2025; Asante-Appiah & Lambert, 2022). Therefore, Equation (2) is presented, which will be tested as a robust measure:
In comparison to Equation (1), in addition to the variable Tarnished Reputation (REPM) replacing the variable Abnormal Idiosyncratic Volatility (AIV), other changes occurred. The robustness test is important because it reduces the possibility of endogeneity problems and also strengthens the relationship between the variables studied. The variables used in the study are shown in
Table 1 below, which identifies the abbreviation used in the equations, the name of the variables, description, calculation, and theoretical basis. It also classifies the variables as dependent, independent, and control variables.
The dependent variable used in this research will be AIV (Abnormal Idiosyncratic Volatility), which captures signs of opportunistic insider trading. Research by Yang, Zhang, and Zhang (2020) found a positive relationship between AIV and the intensity of abnormal gains before earnings release. It also revealed that stocks with negative AIV have a lower information risk. It is also noteworthy that AIV is a measure of information risk based on stock price, because when there are insider trading operations, in addition to increased trading intensity, there is also a change in stock prices. Corroborating this statement, the literature shows that insider trading that occurs before earnings announcements has a greater impact on stock prices, because in an environment of uncertainty, information becomes relevant (Arrow, 1963; Chauhan, Kumar & Chaturvedula, 2016).
Therefore, to construct the AIV variable, the following five steps were necessary:
1. Data collection.
2. Calculation of the daily residual based on the previous year.
3. Sum of the daily residuals for each period of PEA and NEA.
4. Calculation of IVpea and IVnea.
5. Calculation of AIV.
The PEA refers to the five business days before disclosure, and the NEA refers to all days for one year, excluding the eleven days surrounding each disclosure of information. That is, the eleven days consist of the day of disclosure, five business days before, and five business days after the disclosure of the information. As explained in the research data, the information we used as a basis for calculating the AIV was the dates and times of disclosure of quarterly and annual financial statements and material facts. Regarding the time, it was defined that: (i) when the firm disclosed between 00:00 and 10:00 am, the date remained the same; (ii) when the firm disclosed between 10:01 am and 11:59 pm, the date changed to the following day. This occurred because disclosure during the stock exchange trading period can negatively impact investors’ decisions when trading shares.
After collecting the publication dates for each document, each firm, and each year, the NEA and PEA periods were calculated in an Excel spreadsheet. Following this step, more than one million daily regressions were performed, requiring looping programming in the Stata statistical software and expertise in processing the collected data. This amount of regression occurred because, to obtain the daily residual, it was necessary to perform a regression for each day. That is, 3024 days, using daily data from the previous 252 days. For this, the three-factor model (FF-3) of Fama and French (1993) was used, as demonstrated:
where: Rit = Return of firm i on day t; RFt = Risk-free rate, calculated from the 30-day DI Swap; MKT = Difference between the daily returns weighted by the market value of the portfolio on day t; SMB = Return of a portfolio long in stocks with low market capitalization (small) and short in stocks with high market capitalization (large) on day t; HML = Return of a portfolio long in stocks with a high book-to-market ratio and short in stocks with a low book-to-market ratio on day t; ε = Residual of the model referring to portfolio i on day t.
With this information, the idiosyncratic volatility was calculated for the pre-earnings announcement period (IVPEA) and for the days after earnings announcements (IVNEA), using the portfolio return as the logarithm, according to Equation (4), considering that there are 252 trading days during a year, according to the following equations:
where: nPEA = number of days before results announcement. nNEA = number of days after results announcement.
After finding the volatility for the two periods, according to Equations (4) and (5), these two components were subtracted to find the idiosyncratic volatility component that is related to information risk surrounding the disclosure of company earnings, according to the following model:
Considering that AIV refers to the difference in idiosyncratic volatility between periods before information disclosure and periods without disclosure, and that insider trading occurs before a disclosure period, it is believed that AIV can capture evidence of opportunistic insider trading, according to a study by Yang, Zhang, and Zhang (2020). To test the research hypothesis, variables representing aspects of monitoring, oversight, and transparency of corporate governance were included in the model, since, according to the literature on the subject (Ravina & Sapienza, 2010; Tang, Chen & Chang, 2013; Ali & Hirshleifer, 2017; Jacob, 2019; Rahman, Faff & Oliver, 2020; Wu, Sorensen & Sun, 2019; Javakhadze, Pennathur & Silverstein, 2025; Contreras & Marcet, 2021), governance has aspects that can mitigate the opportunistic behavior of insiders.
The variable GC (Corporate Governance) was included in the model because some research indicates that governance mechanisms mitigate opportunistic behavior (Shleifer & Vishny, 1997; La Porta et al., 2000; Ali & Hirshleifer, 2017; Contreras & Marcet, 2021; Javakhadze, Pennathur & Silverstein, 2025). Conversely, adherence to differentiated levels of governance does not influence the improvement of governance practices (Black, Carvalho & Gorga, 2010; Ventura et al., 2024). Furthermore, depending on the culture and legislation, corporate governance mechanisms are not as effective in minimizing opportunism (Ventura et al., 2024; Macagnan, 2025). Given this context, it is believed that a firm assuming obligations that go beyond legal requirements may reduce unethical behavior within the firm.
Additionally, other variables representing monitoring, oversight, and transparency practices were included in the model, considering the behavioral aspect. It is expected that the variables proportion of women on the Board of Directors (PMCA), existence of a Fiscal Council (CF), and size of the audit board (TAMCAUD) will mitigate the indications of opportunistic insider trading, since, according to the literature, these three factors provide greater monitoring within the firm (Dash, 2012; Procianoy & Decourt, 2015; Elgammal, El-Kassar & Messarra, 2018; Borba et al., 2019; Rahman, Faff & Oliver, 2020; Gupta et al., 2020; Jain & Zaman, 2020; Wu, Sorensen & Sun, 2019; Ngo & Le, 2021).
Furthermore, regarding monitoring and oversight, there is the variable of firms audited by Big Four (AUDB4), in which some studies show that audit quality is better when performed by one of the largest firms (Francis & Yu, 2009; Eshleman & Guo, 2014). However, more recent studies show that the Big Four have been providing consulting services to firms involved in scandals (Abid et al., 2018; Donelson et al., 2020; Hasnan et al., 2022; Friedrich & Quick, 2023). Therefore, the relationship between the Big Four and audit viability can be positive, considering that audit quality decreases when they provide consulting services, or the relationship can be negative, since larger audit firms will have more resources to carry out greater oversight and monitoring.
Increasingly, managers are concerned about the risk to the firm’s reputation and how their opportunistic behavior ex ante could cause ex-post reputational damage (Gao, Lisic & Zhang, 2014). In this sense, considering that a firm’s ESG score, which is an intangible asset that helps promote self-discipline among managers (Gao, Lisic & Zhang, 2014), was included as a proxy for the Good Reputation variable (REPB). Considering the firm’s fear of being involved in scandals related to opportunistic insider trading, as well as the concern about the firm’s image in the market, it is believed that firms with ESG scores adopt ethical practices within the company.
Given that greater transparency reduces information asymmetry and, consequently, the insider’s informational advantage, the variable “number of analysts covering the firm” was added to the model as a proxy for Information Quality (IQ). Analyst coverage corresponds to the absolute number of analysts who followed each firm in each period analyzed. Some studies have used this variable as a proxy for information asymmetry between insiders and outsiders in companies without analyst coverage (Hillier, Korczak & Korczak, 2015; Ellul & Panayides, 2018). Therefore, it is expected that IQ will reduce the signs of opportunistic insider trading.
Rahman, Faff, and Oliver (2020) showed that board independence restricts opportunistic insider trading in Australian firms. However, family-owned firms with high shareholding control may dilute board independence, reducing the effectiveness of monitoring (Jaggi & Tsui, 2007). Additionally, it was revealed that in the emerging market of Taiwan, family-owned firms engage in many insider trading transactions (Tang, Chen & Chang, 2013). Furthermore, according to the authors, firms with high ownership and shareholding control tend to leverage their control to manipulate results to engage in insider trading. Given this context, it is expected that Non-Family Firms (NFF) will reduce the likelihood of opportunistic activities and that the Proportion of Family Members on the Board (PFCA) will mitigate monitoring, thereby increasing the possibilities of misconduct for self-interest.
Considering informational asymmetry, several control variables that demonstrate firm information to the market were included in the model. Dividend payment, represented by the Dividend Yield (DY) variable, is directly related to the organization’s future (Pietro Neto, Decourt & Galli, 2011). In this sense, shareholders who trust the firm prefer not to receive dividends so that investments can be made. On the other hand, when shareholders do not see prospects for the firm, they demand to receive the payment (La Porta et al., 2000). Furthermore, Simon, Procianoy, and Decourt (2019) found that low profitability in institutions may be related to a high distribution of dividends, due to the intention of signaling good future results through dividend policy. Thus, a positive relationship between DY and AIV is expected. The following equation was used to calculate Dividend Yield:
where: DY = dividend yield. D = amount of dividends paid per share; and Pt-1 = value of the company’s share on the day before the announcement date.
The variable Tarnished Reputation (REPM) is represented by the number of republications of financial statements. This variable was chosen because firms that exhibit opportunistic behavior in manipulating results and have characteristics of low financial performance are equated with firms with high republication rates (Ali & Hirshleifer, 2017; Javakhadze, Pennathur & Silverstein, 2025; Martins & Ventura Júnior, 2020). Furthermore, the occurrence of republications has been used in several studies as a proxy for misconduct (Ali & Hirshleifer, 2017), thus justifying the use of this variable for the robustness test. Regarding the main model, a positive relationship is expected between REPM and AIV.
The asset’s natural logarithm will be a proxy for the Media Attention (MEDIA) variable, considering that the size of the company influences the attention received from the media, as well as from the most attentive analysts (Hodgson, Seamer & Uylangco, 2020). Another recent study also used firm size as a proxy for the level of media coverage (Asante-Appiah & Lambert, 2022). Furthermore, the literature (Vasconcelos, Gadi & Monte-Mor, 2016; Ali & Hirshleifer, 2017; Borochin, Ghosh & Huang, 2019) indicates that smaller firms are more likely to adopt insider trading policies, as well as to perform greater earnings management. Therefore, it is believed that concern about the firm’s impact on society and the market inhibits managers from engaging in opportunistic actions.
Volatility was chosen to be part of the model because it is related to the uncertainty of the financial market and, consequently, affects investors’ decisions. Volatility is a measure used for risk and constitutes the rate of change in the price of a security at a given moment (Bhowmik & Wang, 2020). Therefore, the higher the volatility, the greater the chance of gain or loss in the short term, increasing the risk of the asset in question. Thus, volatility will be used as a proxy for Leaked Information (IVAZ) and its calculation was performed according to Equation (8):
where, VOLATi,t = Volatility of the closing price of corporation i’s stock in year t. Sd = natural logarithm of (Pd/Pd-1); where d = 1 … n and Pd = closing price of the stock on day d.
= Average of Sd in the year. n = Number of quotation days in the year.
The risk of information leaks is associated with trading involving information that has not yet been disclosed (Kacperczyk & Pagnotta, 2019). That is, an increase in the number of people involved in such activity, or if a firm is about to announce mergers and acquisitions, will increase the risk of information leaks to the public. Therefore, the market will pay closer attention to the company and, therefore, demand higher returns for assuming the legal risk. In this sense, Kacperczyk and Pagnotta (2019) argue that the volatility of asset prices and abnormal trading reflects private information to the public, which subsequently determines the action of carrying out illegal trades with privileged information. Therefore, it is expected that Information Leaks (IVAZ) are directly related to evidence of opportunistic insider trading.
The control variable representing Profitability (RENT) is measured by asset turnover multiplied by net profit margin. It is important to add this variable to the model to control effects related to the firm’s financial performance. Given that managerial opportunism is negatively associated with accounting and performance-based measures of the firm, a negative relationship is expected between profitability and firm misconduct. The results found will be presented, analyzed, and discussed below.