Article

Jorge Lucas Martins da Silva; Janaina Macedo Calvo

DOI: 10.22167/2675-6528-2026M22

Article prepared by the ResumeAI tool, an artificial intelligence solution developed by the Pecege Institute focused on synthesis and writing

Abstract

The study analyzed the influence of specific risk on earnings management in Brazilian publicly traded companies. A quantitative and descriptive approach was adopted, with an econometric character, using data from 259 non-financial companies listed on the Brazilian market between 2019 and 2024. Earnings management was operationalized through discretionary accruals, estimated by the Dechow model, while specific risk was measured by the volatility of the residuals from the asset pricing model. The fixed-effects regression indicated a positive association between specific risk and earnings management. Quantile regression revealed that the influence is heterogeneous and statistically significant at the 0.25 and 0.90 quantiles. Company size and return on assets showed negative associations, while leverage showed a positive and increasing association along the distribution. The market-to-book ratio did not show statistical significance. The results aligned with the assumptions of agency theory and international literature, suggesting that earnings management does not occur uniformly but is conditioned by company-specific factors. The research contributed to the literature by demonstrating that risk heterogeneously affects managers’ behavior, offering implications for investors, credit analysts, and regulators by indicating the need for greater scrutiny of companies with high specific risk.

Keywords: Accruals Discricionários; Capital Market; Quantile Regression.

1. Introduction

Understanding the dynamics between risk and return is fundamental for investment decision-making and corporate management. Portfolio theory, developed by Markowitz (1952), laid the groundwork for constructing investment portfolios, emphasizing diversification as an essential mechanism for optimizing the risk-return relationship. Building on these foundations, the Capital Asset Pricing Model (CAPM), proposed by Sharpe (1964), Lintner (1965), and Mossin (1966), further distinguished between systematic risk and specific risk.

Within the scope of CAPM, systematic risk refers to the portion of an asset’s return volatility that is explained by the fluctuations of the market as a whole, being non-diversifiable. On the other hand, specific risk comprises the volatility component not explained by the market, being theoretically diversifiable. Traditionally, CAPM assumes that investors can eliminate specific risk through the diversification of their portfolios, focusing only on systematic risk as a determinant of expected returns. However, this metric has been increasingly used as a proxy for individual firm risk exposure, reflecting crucial aspects related to the quality of financial information and corporate governance (Mitra, 2016; Datta et al., 2017; Widianingsih et al., 2022).

In markets characterized by informational asymmetry, specific risk transcends the mere condition of diversifiable noise, assuming a relevant role in corporate transparency and the credibility of disclosed financial information (Malagon et al., 2015). Informational asymmetry, a central concept in agency theory, postulates that managers possess privileged information about the company compared to shareholders (Jensen and Meckling, 1976). This informational advantage can be exploited by managers to prioritize their own interests, often to the detriment of shareholders. One of the manifestations of this exploitation is earnings management (EM), defined as the application of accounting discretion to manipulate financial information, aiming to achieve contractual, regulatory, or market objectives (Healy and Wahlen, 1999).

Earnings management, by shaping the firm’s perceived performance, can alter external volatility and consequently influence specific risk. Firms facing higher specific risk may be incentivized to use discretionary accruals to smooth their earnings (Datta et al., 2017) or to create an artificial perception of financial stability. This practice can distort stock pricing in the market and, paradoxically, increase the specific risk perceived by investors (Widianingsih et al., 2022), creating a feedback loop between risk and accounting practices.

Despite the relevance of the relationship between specific risk and earnings management, international literature has explored this theme more comprehensively than the national context. In Brazil, an emerging market, several institutional factors can influence the nature of this relationship. Informational inefficiencies, structural imperfections, the heterogeneity of corporate governance mechanisms, and the low liquidity of certain assets are characteristics that can modify how specific risk interacts with earnings management, making a specific analysis for the Brazilian market imperative.

Given this scenario and the existing gap in the national literature, this study is justified by the need to understand the particularities of the Brazilian capital market and its impacts on the accounting behavior of companies. Thus, the central objective of this research is to examine the influence of specific risk on the levels of earnings management in publicly traded companies in Brazil, contributing to the theoretical and practical advancement on the subject and providing results aligned with the country’s institutional specificities.

2. Material and Methods

The research was characterized as quantitative and descriptive, with an econometric approach. The central objective was to examine the influence of specific risk on earnings management levels in publicly traded companies in Brazil. The study population comprised companies listed on the Brasil, Bolsa e Balcão (B3) with standardized financial statements and data available in the Refinitiv Eikon® database. Financial institutions and companies without complete data for variable measurement were excluded. The final sample consisted of 259 non-financial companies. The research time frame covered the period from 2019 to 2024.

Financial and market data were collected from the Refinitiv Eikon® database and the B3 portal. Earnings management (EM), the dependent variable, was operationalized through discretionary accruals (DA), following the most common method in the literature. Total accruals were obtained by the difference between net income and operating cash flow (Hribar and Collins, 2002). Non-discretionary accruals were estimated by the model of Dechow et al. (1995), adjusted for variations in revenues, accounts receivable, and fixed assets. The absolute values of discretionary accruals were used to capture the intensity of the phenomenon. Specific risk, the independent variable, was measured by the volatility of the residuals from the Capital Asset Pricing Model (CAPM), according to the literature (Ang et al., 2006). Adjusted stock prices and monthly returns data were obtained from the Refinitiv Eikon® database. Market return was represented by the Ibovespa index and the risk-free rate by the Interbank Deposit Certificate (CDI), both obtained from the B3 portal. The estimation of each company’s specific risk was performed through the regression of the CAPM model. The volatility of the residuals was measured by calculating the monthly standard deviation, annualized by multiplying by the square root of 12.

To isolate the effects of the test variable, control variables were included: firm size, measured by the natural logarithm of total assets (Chang et al., 2015), and leverage, measured by the ratio of liabilities to total assets (Usman et al., 2022). The Market-to-Book (MTB) ratio was calculated as the market value of equity relative to its book value (Usman et al., 2022). Profitability (Return on Assets – ROA) was measured by the ratio of net income to total assets (Flores and Sampaio, 2018). The main regression model was estimated to analyze the relationship between earnings management (dependent variable) and specific risk (independent variable), controlling for the variables of size, profitability, leverage, and market-to-book ratio. Data analysis began with descriptive statistics.

The Variance Inflation Factor (VIF) test was performed to check for the absence of multicollinearity. For the definition of panel modeling, the Chow, Hausman, and Breusch-Pagan tests were applied. The presence of heteroscedasticity in the residuals, indicated by the Breusch-Pagan test, led to the use of robust standard errors in the fixed-effects panel regression. Considering the asymmetric distribution of earnings management and the rejection of normality by the Shapiro-Wilk test, a quantile regression was applied. This technique was used to examine different forms of heterogeneity in the sample distribution (Ramdani and Witteloostuijn, 2010; Usman et al., 2022). Quantile regression was applied at the 0.10, 0.25, 0.50, 0.75, and 0.90 quantiles (Chang et al., 2015). To mitigate the influence of outliers, the sample was subjected to winsorization at the 1% and 99% levels.

3. Results and Discussion

The analysis of the collected data revealed a series of important characteristics regarding earnings management and specific risk in Brazilian publicly traded companies. Initially, descriptive statistics indicated that earnings management (EM) presented a mean of 0.1089, higher than the median of 0.0665, accompanied by a positive skewness of 7.605 and a high kurtosis of 107.157. These values suggest that the occurrence of earnings management is concentrated in a smaller group of companies, which exhibit more extreme levels of this practice, with high data dispersion. Specific risk, in turn, showed similar behavior, with a mean of 0.4204 and a median of 0.3582, in addition to positive skewness of 2.183 and kurtosis of 6.076, indicating that high risk is also concentrated in a specific subset of observations within the sample.

Regarding the control variables, firm size showed an almost symmetric distribution, with close mean and median and low kurtosis, suggesting a more homogeneous distribution in the sample. Return on Assets (ROA) exhibited negative skewness and high kurtosis, which can be attributed to the presence of firms with negative performance, i.e., losses, in the left tail of the distribution. Leverage and the Market-to-Book ratio (MTB) showed positive skewness and high kurtosis, indicating, respectively, that high levels of debt and greater growth expectations are concentrated in a smaller portion of the analyzed firms. These distribution characteristics are crucial for the choice and interpretation of econometric models, especially quantile regression.

The analysis of Pearson correlations provided a preliminary insight into the interdependencies between the variables. A moderate positive correlation was observed between specific risk and leverage (0.37), and between specific risk and earnings management (0.21). These findings suggest that companies with higher risk exposure tend to exhibit higher levels of indebtedness and, concomitantly, a greater propensity for earnings management. In contrast, moderate negative correlations were identified between company size and specific risk (-0.35), as well as between ROA and specific risk (-0.28). This indicates that larger and more profitable companies tend to be associated with lower risk, which may reflect greater stability and less uncertainty perceived by the market.

The temporal trajectory of average annual earnings management in the sample, between 2019 and 2024, revealed heterogeneous behavior. An increase in EM was observed starting in 2019, reaching a peak in 2020, with an average of 0.166. This increase can be associated with economic instabilities and uncertainties arising from the Covid-19 pandemic, a result that aligns with the findings of Flores et al. (2023) for the Brazilian market in the same period. After the 2020 peak, there was a gradual decrease in average EM levels until 2023, followed by a new increase in 2024, reaching 0.113. This temporal dynamic suggests that earnings management is sensitive to macroeconomic shocks and periods of instability.

For the validation of the econometric model, specification tests were applied. The Variance Inflation Factor (VIF) test indicated the absence of multicollinearity, with the highest VIF recorded for leverage (1.45), followed by ROA (1.35), size (1.15), and MTB (1.06), all below the limit of 5, which ensures that the coefficients are not inflated by excessive correlations between the explanatory variables (Greene, 2018). Subsequently, panel data adequacy tests were performed: the Chow test rejected the null hypothesis that the Pooled OLS model would be adequate, and the Hausman test rejected the hypothesis of consistency of the random effects panel. These results indicated that the fixed effects model was the most appropriate for data analysis. Additionally, the Breusch-Pagan test pointed to the presence of heteroscedasticity in the residuals, which justified the use of robust standard errors for the estimations (Fávero and Belfiore, 2017; Greene, 2018).

The results of the fixed-effects panel regression, with robust standard errors, indicated a positive and marginally significant association between specific risk and earnings management, with a coefficient of 0.0422 (p < 0.10). This finding suggests that the higher a company’s specific risk, the greater its propensity to engage in earnings management practices. Such a result is in line with the international literature exploring this relationship (Mitra, 2016; Chang et al., 2015; Datta et al., 2017; Widianingsih et al., 2022). The company size variable, in turn, showed a negative and statistically significant association (-0.0503, p < 0.01), corroborating the hypothesis that larger companies are subject to greater scrutiny by auditors, regulators, and other stakeholders, which tends to inhibit earnings management (Mitra, 2016). The other control variables, leverage, ROA, and MTB, did not show statistical significance in this fixed-effects model.

Considering the asymmetric and leptokurtic distribution of earnings management, as evidenced by descriptive statistics and histograms, and the low explanatory power of the Within R² of the fixed-effects model, which may be limited by the violation of classical distribution assumptions, the Shapiro-Wilk normality test was performed. The result of this test, with a p-value lower than 0.05, led to the rejection of the null hypothesis of normality of the residuals’ distribution (Fávero and Belfiore, 2017). This finding justified the application of quantile regression, a more suitable approach to analyze the influence of explanatory variables at different points of the earnings management distribution, without requiring assumptions of Gaussian error distribution and allowing the capture of heterogeneities in the sample (Ramdani and Witteloostuijn, 2010; Usman et al., 2022).

Quantile regression was applied to the quantiles 0.10, 0.25, 0.50, 0.75, and 0.90, revealing that the influence of specific risk on earnings management is not homogeneous across the distribution. A positive and statistically significant association was observed in the lower quantiles (τ = 0.25, with a coefficient of 0.0167 and p < 0.05) and in the upper extreme quantiles (τ = 0.90, with a coefficient of 0.1038 and p < 0.01). In the other quantiles (0.10, 0.50, and 0.75), the relationship was not statistically significant. The disparity in the magnitude of the significant coefficients indicates that companies with higher levels of earnings management (top of the distribution) are more sensitive to specific risk, suggesting that the intensive use of discretionary accruals may be a response to the high volatility of specific risk. This pattern is consistent with the classification by Usman et al. (2022), which indicates that companies in the upper quartiles of GR use this practice more sharply in response to adverse economic pressures, such as risk.

The quality of the model fit, measured by the Pseudo R² of quantile regression, showed progressive growth along the distribution, ranging from 0.0196 in the lower quantile (τ = 0.10) to 0.1168 in the upper quantile (τ = 0.90). This increase in explanatory power suggests that the explanatory variables have greater statistical adherence and capacity to explain earnings management in higher ER scenarios. Such validation reinforces the relevance of choosing quantile regression to capture the heterogeneous and divergent behavior of earnings management, which would not be adequately revealed by models that assume a constant mean for the entire distribution, such as the fixed-effects model.

The heterogeneity of the effects of specific risk on earnings management can be interpreted from two perspectives that, although distinct, are convergent. The first perspective suggests a feedback loop, where an increase in specific risk intensifies the incentive for earnings management. In turn, earnings management can deteriorate information quality, introducing noise into stock prices and, paradoxically, again increasing the specific risk perceived by investors (Mitra, 2016; Datta et al., 2017). In this context, earnings management would be used to smooth investors’ perception of volatility, seeking artificial stability.

The second perspective suggests that specific risk can be a reflection of imperfections in both corporate investments and in investors’ stock “mispricing” (Malagon et al., 2015). Deviations between market value and company fundamentals can occur when disclosed information does not adequately represent the company’s real conditions (Widianingsih et al., 2022). In both scenarios, the obtained results corroborate that, in higher risk environments, managers tend to intensify earnings management, either to correct internal investment distortions or to mitigate informational asymmetries in the market, seeking to influence the external perception of performance and stability.

Regarding the control variables in quantile regression, firm size presented negative coefficients in all quantiles, being statistically significant at 5% or 1% in most of the distribution, except in the lower quantile (τ = 0.10). This reinforces the idea that larger firms tend to show a lower propensity for earnings management, probably due to greater scrutiny by stakeholders, which increases transparency and reduces managerial discretion (Mitra, 2016). Leverage, on the other hand, exhibited positive and significant coefficients, with increasing values throughout the entire distribution of quantiles. This result indicates that higher levels of debt are associated with greater earnings management (Mitra, 2016), possibly driven by the pressure of debt covenants that motivate managers to manipulate earnings to avoid contractual violations.

The return on assets (ROA) showed negative and significant coefficients in the quantiles 0.10, 0.25, 0.50, and 0.75. This finding suggests that companies with higher profitability tend to exhibit a lower propensity for earnings management (Usman et al., 2022), as they can meet market expectations through organic operational performance. However, in the extreme quantile (τ = 0.90), the relationship between ROA and earnings management lost statistical significance, indicating that at very high levels of EM, managers’ discretionary behavior may become dissociated from the company’s actual operational performance. The market-to-book ratio (MTB) did not show statistical significance in any of the quantiles, a result that aligns with the literature (Mitra, 2016; Usman et al., 2022), suggesting that this metric is not a determining factor of earnings management in the analyzed context.

In summary, the results of this research demonstrate that specific risk exerts a positive and heterogeneous influence on earnings management in publicly traded Brazilian companies. This influence is more pronounced in companies that already practice higher levels of earnings management, as evidenced by quantile regression. Control variables, such as size and return on assets, tend to reduce earnings management, while leverage increases it, especially at higher levels of indebtedness. These findings align with the assumptions of agency theory, indicating that managers adjust their accounting behavior in response to specific incentives and pressures from the corporate and market environment, confirming that earnings management does not occur uniformly but is conditioned by factors intrinsic and extrinsic to the companies.

4. Conclusion

This study examined the influence of specific risk on earnings management in publicly traded companies in Brazil. It was found that specific risk showed a positive association with earnings management, as indicated by the fixed-effects regression. However, the quantile regression analysis revealed that this influence is not homogeneous, being statistically significant and more pronounced in companies with moderate and high levels of earnings management. It was observed that company size and return on assets were associated with a lower propensity for earnings management, while leverage demonstrated a positive and increasing association along the distribution. The market-to-book ratio did not show statistical significance. These findings align with the assumptions of agency theory, suggesting that managers’ behavior is adjusted in response to incentives and pressures from the corporate and market environment.

The main contribution of this work lies in demonstrating that specific risk heterogeneously affects managers’ behavior in the Brazilian context, broadening the understanding of earnings management dynamics in emerging markets. The results offer relevant practical implications for investors, credit analysts, and regulators, by indicating the need for greater scrutiny of companies with high specific risk, as well as the strengthening of disclosure practices that can mitigate informational asymmetries and reduce potential pricing errors. Despite the contributions, some inherent limitations of the study are acknowledged, such as the relatively small set of control variables and the restrictions of the “proxies” used to measure earnings management and specific risk. Additionally, the methodological design did not address possible endogeneity problems, such as simultaneity and reverse causality. For future studies, it is suggested to use alternative metrics for the variables of interest, conduct segmented analyses by sector, expand the temporal base, and include other control variables. It is also recommended to apply more advanced econometric models capable of mitigating potential simultaneity and reverse causality biases, deepening the robustness of the results.

Bibliographic References

Ang, A.; Hodrick, R.J.; Xing, Y.; Zhang, X. 2006. The cross-section of volatility and expected returns. The Journal of Finance 61(1): 259-299.

Chang, S.; Wang, T.; Chiu, A.A.; Huang, S.Y. 2015. Earnings management and idiosyncratic risk – evidence from the post-Sarbanes-Oxley Act period. Investment Management and Financial Innovations 12(2): 117-126.

Datta, S.; Iskandar-Datta, M.; Singh, V. 2017. The impact of idiosyncratic risk on accrual management. International Journal of Managerial Finance 13(1): 70-90.

Dechow, P.M.; Sloan, R.G.; Sweeney, A. 1995. Detecting earnings management. The Accounting Review 70(2): 193-225.

Flores, E.; Sampaio, J.O. 2018. The extension of the modified Jones model with control variables: empirical findings from the Brazilian capital market. International Journal of Auditing Technology 4(1): 36-48.

Flores, E.S.; Sampaio, J.O.; Beiruth, A.X; Brugni, T.V. 2023. Earnings management during the COVID-19 crisis: evidence from the Brazilian and American capital markets. Journal of Accounting in Emerging Economies 13(4): 760-783.

Fávero, L.P.; Belfiore, P. 2017. Manual de Análise de Dados: Estatística e Modelagem Multivariada com Excel®, SPSS® e Stata®. Gen LTC, Rio de Janeiro, RJ, Brasil.

Greene, W.H. 2018. Econometric Analysis. 8ed. Pearson, New York, NY, USA.

Healy, P.M.; Wahlen, J.M. 1999. A review of the earnings management literature and its implications for standard setting. Accounting Horizons 13(4): 365-383.

Hribar, P.; Collins, D.W. 2002. Errors in estimating accruals: Implications for empirical research. Journal of Accounting Research 40(1): 105-134.

Jensen, M.; Meckling, W. 1976. Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3(4): 305-360.

Lintner, J. 1965. Security prices, risk, and maximal gains from diversification. The Journal of Finance 20(4): 587-615.

Malagon, J.; Moreno, D.; Rodríguez, R. 2015. The idiosyncratic volatility anomaly: Corporate investment or investor mispricing?. Journal of Banking & Finance 60: 224-238.

Markowitz, H. 1952. Portfolio selection. The Journal of Finance 7(1): 77-91.

Mitra, R.K. 2016. The association between earnings quality and firm-specific return volatility: Evidence from Japan. Review of Accounting and Finance 15(3): 294-316.

Mossin, J. 1966. Equilibrium in a capital asset market. Econometrica 34(4): 768-783.

Ramdani, D.; Witteloostuijn, A. 2010. The impact of board independence and CEO duality on firm performance: A quantile regression analysis for Indonesia, Malaysia, South Korea and Thailand. British Journal of Management 21(3): 607-626.

Sharpe, W.F. 1964. Capital asset prices: A theory of market equilibrium under conditions of risk. The Journal of Finance 19(3): 425-442.

Usman, M.; Nwachukwu, J.; Ezeani, E. 2022. The impact of board characteristics on the extent of earnings management: conditional evidence from quantile regressions. International Journal of Accounting & Information Management 30(5): 600-616.

Widianingsih, Y.P.N.; Setiawan, D.; Aryani, Y.A.; Gantyowati, E. 2022. Accrual management and firm-specific risk. International Journal of Financial Studies 10(4): 111.

Article originating from the Final Project of Specialization in Finance and Controllership of the MBA USP/Esalq

_{Learn more about the course; click here:}