Influence of Ownership Structure on the Debt Level and Efficiency of Electricity Companies

Abstract

Corporate ownership structure has been extensively studied as a complementary indicator of corporate performance; however, there is a lack of research on specific sectors. This research aims to understand the effects of ownership structure on debt levels and the efficiency of companies in the electricity sector. The efficiency of electricity companies can help reduce dependence on fossil fuels. The analysis covers 7777 observations of 38 companies from 1996 to 2023, including all publicly traded electricity companies on the Brazilian stock exchange (B3). The method is based on a panel data regression model with statistical tests to support the results. The results were robust and significant. Regarding efficiency, ownership structure revealed a negative relationship (−0.107); however, as company capital becomes more dispersed, it ceases to influence company efficiency. In terms of debt levels, the relationship was positive (0.095) and, despite the dispersion of capital among the five largest common shareholders (0.103), continues to exert influence. This research contributes to the scientific literature by confirming relationships and providing evidence between new and previously unexplored variables specific to the electricity sector. It is expected to create a benchmark for future analyses, highlighting the importance of ownership structure in the performance of electricity sector companies, which can contribute to improving the management of these companies, making them more competitive and sustainable.

1. Introduction

Agency conflicts occur in corporate governance when the interests of agents (such as administrators or managers) are not aligned with the interests of principals (usually the shareholders or owners) [1]. As companies develop and grow, the separation between ownership and control occurs almost naturally, which can compromise the company’s performance. While more concentrated structures may alleviate this problem, they can also overshadow the interests of minority shareholders; there exists a trade-off [2]. As concentration increases, this conflict tends to diminish; however, a new conflict arises, that of personal appropriation by controlling shareholders at the expense of company performance, which affects minority shareholders.

These relationships gained prominence starting in 1976 with the first studies [1] on the influence of ownership structure on company performance. Since then, studies have been replicated in different locations and contexts [3,4,5]. The importance of ownership structure was consolidated in these studies; however, the analyses were conducted globally [6,7], across the entire economy, without specific sectoral analyses. This research analyzed the ownership structure specific to the Brazilian electricity sector, filling a gap in the literature. Understanding how it influences the Brazilian electricity sector can contribute to improving the performance and efficiency of electricity companies, making them more competitive and sustainable. This can indirectly contribute to improving the efficiency of electricity companies and help reduce dependence on fossil fuels. Another contribution of this research is to provide information for the development of public policies by sector regulatory bodies. Understanding how the ownership structure interacts within the sector can inform managers’ decision-making. Everyone—government, managers, users, and society as a whole—can benefit from the results of this research; an efficient electricity sector is beneficial to all stakeholders; everyone consumes electricity, and improving efficiency can help reduce pollution and contribute to reducing pollutant emissions.

The Brazilian electricity sector encompasses the generation, transmission, distribution, and commercialization of electric power. This energy chain is fundamental for sustaining the country’s infrastructure and economic development. According to the 2022 National Energy Balance (BEN), the Brazilian energy matrix is predominantly composed of renewable sources, reaching 78%. The global average, according to data from the International Energy Agency (IEA), is around 29% [8].

The electricity sector is crucial for the economy, as it enables the production and consumption of energy by businesses and individuals. Given its significance, in 1998, B3 launched the Electric Energy Index (IEE B3), aiming to serve as an indicator of the average performance of the more negotiable and representative assets in the electricity sector. As of April 2024, the companies included in the index represented a market value of $75 billion [9].

Studies have demonstrated the relationship between ownership structure and efficiency [10], although without sectoral analysis, which is the distinctive feature and objective of this study. This research innovates and differentiates itself by examining the relationship between ownership structure and other company indicators through a sectoral analysis of the electricity sector using econometric techniques.

The research problem is presented in the form of a question: Does ownership structure influence the indebtedness and efficiency of companies in the electricity sector? In this context, the objective of this research is to investigate whether ownership structure affects the efficiency and indebtedness of electric sector companies in Brazil. The originality of the research lies in contributing to the literature with empirical results on the effects of ownership structure on the electricity sector. Stakeholders and regulators can benefit from the information generated in policymaking and decision-making for the sector; society as a whole can benefit from this information, considering that an efficient electricity sector would help reduce emissions and pollution.

Studies have associated governance/sustainability with ownership concentration [11], while others have analyzed the link between oligopolies and ownership structure [12]. Research examining the perceptions of economic actors is found relatively frequently [13]. In contrast, this study breaks new ground, filling this gap in the literature by analyzing the impact on key sectors of the economy, specifically the electricity sector, considering that the efficiency of this sector can contribute to sustainability, providing information to regulators and stakeholders in decision-making.

The results confirmed the influence of ownership structure on companies in the electricity sector. Regarding efficiency, the relationship was negative; however, as company capital becomes more dispersed, it ceases to influence company efficiency. In terms of debt levels, the relationship was positive and, despite the dispersion of capital, continues to exert influence. These data provide insights to managers of these companies and policymakers for the sector, especially in concentrated companies, where this structure negatively affects company efficiency.

The article was structured into the following sections: Introduction, where the research was presented; followed by the Literature Review, in which the bibliographical references that guided the research were discussed; Methodology, where the techniques used were addressed; Results and Discussion, in which the research findings were presented and discussed; Conclusions, in which the implications and suggestions for future research are commented; and finally, the Appendices and References.

2. Literature Review

2.1. Theoretical Background

The theoretical framework is based on the theory of Corporate Governance [2] and Agency Conflict [1], which state that conflict arises when the interests of the contracting party (principal) and the contracted party (agent) diverge. The principal is the one who holds the ownership (as shareholders), while the agent is the one who exercises control and management of a company or fund. The conflict arises because the agent’s decisions may not benefit the principal, generating costs for the organization. Authors have examined the causes and consequences of ownership concentration for companies [14]. Different studies have found varying results regarding the effects of ownership concentration, sometimes showing a positive, sometimes negative relationship. These differences are attributed to the methodologies adopted by each researcher, the periods of analysis, and the sectors studied. The literature review did not identify specific studies on ownership structure for the electricity sector, which reinforces the contribution of this research to the literature by filling a gap in such an important sector as the electricity sector.

In general, research demonstrates a negative relationship between ownership concentration and performance, demonstrating a tendency toward the expropriation of minority shareholders’ interests. More dispersed management tends to be more professional, reducing inefficiencies and the dominance of companies by small groups. At the same time, studies indicate that ownership concentration increases monitoring and reduces agency problems [1].

2.2. Previous Studies and Hypothesis Development

Among the studies conducted, they analyze the relationship between ownership structure and its global influence on companies, interpreting the effects on debt and performance as synonymous with efficiency, sometimes positive, sometimes negative. Hamadi [15], when studying ownership structure in Belgium, found a negative relationship between ownership structure and efficiency, indicating that the lack or weakness of shareholder protection can negatively influence company performance. A similar result was found in Spain by [16]. In Russia, the authors of [17] studied the effects of concentrated ownership on the performance of privatized companies. The results corroborated the thesis that large block ownership is negatively associated with investment and company performance.

Although the effects vary across Europe, ref. [18] concluded that companies without a controlling shareholder tend to outperform their national peers. In contrast, ref. [19] found that concentrated ownership structures increase the performance of companies in IPOs analyzed in the United Kingdom and France. Ref. [20] focused their study on the role of controlling shareholders in US company acquisitions. This influence of controlling shareholders was reflected in the nature of diversification, where ownership concentration had a negative relationship with the number of acquisitions, demonstrating that dispersion is beneficial to this process.

In Asia, the authors of [21] found that in difficult times, non-concentrated structures are better, revealing a negative relationship between ownership concentration and efficiency. Ref. [22], studying publicly traded companies in China, concluded that corporate efficiency is negatively related to state ownership, with a negative relationship between ownership structure and firm efficiency.

Good corporate governance practices maximize value creation for shareholders [2]. Ownership concentration can directly influence this value creation for shareholders, which is linked not only to the concentration of ownership itself but also to the legal origin of the country [23].

Sonza e Kloeckner [24] studied the impact of ownership structure on the efficiency of small and large Brazilian companies listed on B3. Viana et al. [25] investigated ownership concentration and performance in Brazilian companies during periods of initial public offerings. Similarly, ref. [26] researched the relationship between corporate governance, ownership structure, and company performance in Brazil. Despite the absence, as already mentioned, of specific studies on the Brazilian electricity sector, the hypotheses developed were that the electricity sector behaves similarly to the results of other studies carried out in Brazil, cited previously, in which national companies presented a negative relationship between shareholder concentration and performance, due to the greater possibility of expropriation of minority shareholders and the lower professionalization in the company. The alternative hypothesis would be that concentrated ownership structures are positively related to efficiency, as they reduce agency problems through greater monitoring.

3. Methodology

To achieve the objectives, annual data from Brazilian companies listed on B3 (the Brazilian stock exchange) were collected and analyzed for the period from 1996 to 2023. The collected data encompass all companies listed on B3 in the electric power sector (7777 observations from 38 companies; see

Table A1. Code of companies in the electricity sector traded on the Brazilian stock exchange.

AESB3	CPFE3	GPAR3
AESO3	CPLE3	LIGT3
AFLT3	CSRN3	LIPR3
ALUP3	EGIE3	NEOE3
AURE3	EKTR3	REDE3
CBEE3	ELET3	RNEW3
CEBR3	EMAE3	SAEN3
CEEB3	ENEV3	TAEE3
CEED3	ENGI3	TRPL3
CGEE3	ENMT3	UPKP3
CLSC3	EQMA3B
CMIG3	EQPA3
COCE3	EQTL3
COMR3	GEPA3

in Appendix A). Data collection was entirely conducted through the Economática^® database (https://economatica.com/, accessed on 30 August 2025), and the analysis was performed using Stata^® software version 11.

The regression model was based on unbalanced panel data since there are no observations for all variables across all periods, and the data combine cross-sectional and time-series elements. There was a variation in the number of companies active on B3 during the analysis period, with some entering the stock market recently, others privatizing, and some ceasing to exist, alongside mergers and acquisitions contributing to the unbalanced panel. Balanced panels have the same unit over time, while in unbalanced panels, a unit may enter the database and exit before completing the observation period [27].

Two regression models were generated for the following dependent variables: Return on Equity (ROE) used as a proxy for company performance and Total Debt (ET) measured by the ratio of net income to equity. The independent/explanatory variables will be the percentage of ownership concentration (ownership structure of the primary ordinary shareholder and ownership structure of the five ordinary shareholders). To analyze the separate influences, as the controlling capital becomes dispersed, and considering the correlation between these two variables, they were examined in separate regressions, resulting in two regressions for each dependent variable (ET and ROE). The additional explanatory variables are detailed in the following models. Departing from the model developed by [28] and adapted for this work, the following econometric models were arrived at:

$${ET}_{it}=\alpha +{\beta }_R{LnR}_{it}+{\beta }_T{T}_{it}+{\beta }_C{C1}_{it}+{\beta }_D{LnD}_{it}+{\beta }_{PCT}{PCT}_{it}+{\beta }_L{LnL}_{it}+{\beta }_{ROA}{LnROA}_{it}+{\epsilon }_{it}$$

(1)

$${ET}_{it}=\alpha +{\beta }_R{LnR}_{it}+{\beta }_T{T}_{it}+{\beta }_C{C5}_{it}+{\beta }_D{LnD}_{it}+{\beta }_{PCT}{PCT}_{it}+{\beta }_L{LnL}_{it}+{\beta }_{ROA}{LnROA}_{it}+{\epsilon }_{it}$$

(2)

$${ROE}_{it}=\alpha +{\beta }_R{LnR}_{it}+{\beta }_T{T}_{it}+{\beta }_C{C1}_{it}+{\beta }_D{LnD}_{it}+{\beta }_{PCT}{PCT}_{it}+{\beta }_L{LnL}_{it}+{\beta }_{ROA}{LnROA}_{it}+{\epsilon }_{it}$$

(3)

$${ROE}_{it}=\alpha +{\beta }_R{LnR}_{it}+{\beta }_T{T}_{it}+{\beta }_C{C5}_{it}+{\beta }_D{LnD}_{it}+{\beta }_{PCT}{PCT}_{it}+{\beta }_L{LnL}_{it}+{\beta }_{ROA}{LnROA}_{it}+{\epsilon }_{it}$$

(4)

where ${ET}_{it}$ is Total debt; $\alpha$ is Model intercept; ${LnR}_{it}$ is Natural logarithm of gross revenue; ${C5}_{it}$ is Percentage of ownership concentration of the five largest ordinary shareholders; ${C1}_{it}$ is Percentage of ownership concentration of the largest ordinary shareholder; $T_{it}$ is Tangibility of assets; ${LnD}_{it}$ is Natural logarithm of gross debt; ${PCT}_{it}$ is Participation of third-party capital; ${LnL}_{it}$ is Natural logarithm of net income; ${LnROA}_{it}$ is Natural logarithm of ROA; ${ROE}_{it}$ is Return on equity; and ${\epsilon }_{it}$ is residual vector.

The subscripts i and t represent company i at time t. Equations (1) and (2) represent the model for debt in relation to C1 and C5. Equations (3) and (4) represent the model measuring the influence of C1 and C5 on ROE, respectively.

Table 1. Summarizes the variables used and their main information.

Category	Variable	Symbol	Definition	Recent References That Have Used Similar Proxies	Source of Each Variable
Variables explained	Return on Equity	ROE	used as a proxy for company performance, ratio between Net Income and Equity	Sonza e Kloeckner [11]; Sonza e Kloeckner [24]; Viana et al. [25]	Economática® database (https://economatica.com/)
	Total Debt	ET	ratio between Total Gross Debt and Total Assets	Sonza e Kloeckner [11]; Sonza e Kloeckner [24]; Viana et al. [25]	Economática® database (https://economatica.com/)
Explanatory variables	Percentage of ownership concentration of the largest shareholders	C1	Percentage of shareholding control of the main ordinary shareholder	Sonza e Kloeckner [11]; Sonza e Kloeckner [24]; Viana et al. [25]; Silveira [26]	Economática® database (https://economatica.com/)
	Percentage of ownership concentration of the five largest shareholders	C5	Percentage of shareholding control of the five common shareholders	Sonza e Kloeckner [11]; Sonza e Kloeckner [24]; Viana et al. [25]; Silveira [26]	Economática® database (https://economatica.com/)
	Gross Revenue	RB	Natural Logarithm of Gross Revenue	Sonza e Kloeckner [11]; Sonza e Kloeckner [24]; Viana et al. [25]; Silveira [26]	Economática® database (https://economatica.com/)
	Tangibility of Assets	TANG	Ratio between Fixed Assets and Total Assets	Sonza e Kloeckner [11]; Sonza e Kloeckner [24]; Viana et al. [25]; Silveira [26]	Economática® database (https://economatica.com/)
	Gross Debt	DB	Natural Logarithm of Gross Debt	Sonza e Kloeckner [11]; Sonza e Kloeckner [24]; Viana et al. [25]	Economática® database (https://economatica.com/)
	Participation of third-party capital	PCT	Gross Debt to Equity Ratio	Sonza e Kloeckner [11]; Sonza e Kloeckner [24]; Viana et al. [25]	Economática® database (https://economatica.com/)
	Net Profit	LL	Natural Logarithm of Net Profit	Sonza e Kloeckner [11]; Sonza e Kloeckner [24]; Viana et al. [25]	Economática® database (https://economatica.com/)
	Return on Assets	ROA	Natural Logarithm of the Ratio of Net Income to Total Assets	Sonza e Kloeckner [11]; Sonza e Kloeckner [24]; Viana et al. [25]	Economática® database (https://economatica.com/)

summarizes the variables used and their main information. The source for all variables was the Economática^® database (https://economatica.com/, accessed on 30 August 2025).

The values for gross revenue, gross debt, and net income were all collected in U.S. dollars (USD) to avoid inflation effects considering the long analysis period. These same variables (gross revenue, gross debt, and net income) were also winsorized at 5%. Winsorization preserves the influence in the data without compromising the analysis; it allows for the smoothing of potential outliers by replacing them with upper or lower limits [29]. This avoids the definitive loss of extreme points while adjusting them to values within acceptable limits [30].

To choose the appropriate panel data model, Lagrange Multiplier/LM test (Breusch and Pagan) were conducted for random effects versus simple regression. F-statistics (Chow test) compared fixed effects versus simple regression. The Hausman test was used for fixed versus random effects [31].

The next step was to identify high correlation values through the correlation matrix, where values equal to or above 0.8 are considered high, and through the Variance Inflation Factor (VIF) test, where values above 5 indicate the presence of multicollinearity [32].

The presence of heteroscedasticity was checked using White’s [33] statistical test modified for panel data [34]. For autocorrelation of the residuals, the test developed by [35], based on the work of [36], was used. In the presence of heteroscedasticity and autocorrelation, it can be corrected using the Feasible Generalized Least Squares/FGLS procedure [37]. Stationarity was identified through ADF tests adapted for panel data [38,39].

4. Results and Discussion

The descriptive statistics are presented in

Table 2. Descriptive statistics of the companies in the electric sector listed on B3 from 1996 to 2023.

Variable	Number of Observations	Mean	Median	Minimum	Maximum	Standard Deviation
ET	820	0.293	0.300	0.000	3.243	0.236
ROE	820	0.364	0.153	0.000	70.559	2.604
C1	648	0.684	0.670	0.089	1.000	0.232
C5	682	0.894	0.933	0.277	1.000	0.132
RB	763	13.637	13.755	10.790	15.724	1.333
TANG	822	0.305	0.246	0.000	1.000	0.295
DB	759	13.221	13.494	9.398	15.791	1.611
PCT	820	2.067	0.841	0.000	633.909	22.298
LL	821	11.243	11.517	7.988	13.626	1.603
ROA	822	0.076	0.053	0.000	1.735	0.110

ET—Total debt; ROE—Return on equity; C1—Percentage of ownership concentration of the largest common shareholder; C5—Percentage of ownership concentration of the five largest common shareholders; RB—Gross revenue; TANG—Tangibility; DB—Gross debt; PCT—Participation of third-party capital; LL—Net income; ROA—Return on assets.

. As can be observed, after applying the natural logarithm and winsorizing at 5%, the variables Gross Revenue (RB), Gross Debt (DB), and Net Income (LL) showed means and medians that are very close, indicating consistency in the data.

On average, 29% of the companies’ assets would be committed to covering the total of their debts. The return on assets averages only 7%, corresponding to the electric power sector, which requires significant capital investments. The efficiency of management, measured by return on equity, which assesses a company’s ability to generate value for the business and investors based on the resources the company possesses, was 36%, corroborating the significant return that the sector provides to shareholders. The ownership concentration, represented by the control of the largest and five largest ordinary shareholders, averaged 68% and 89%, indicating a high concentration in the sector and a low free float (the percentage of a company’s shares available for trading on the stock market) despite C1 showing a standard deviation of 23%, indicating significant variability in control exercised by the largest shareholder. The average tangibility of 30% indicates that a large portion of the electric power sector companies’ assets are concentrated in fixed assets, configuring real guarantees that can be used in indebtedness.

In

Table 3. Correlation matrix of the companies in the electric sector listed on B3 from 1996 to 2023.

	ET	ROE	C1	C5	RB	TANG	DB	PCT	LL	ROA
ET	1.000
ROE	−0.016	1.000
C1	−0.088	0.001	1.000
C5	−0.124	0.057	0.597	1.000
RB	0.085	−0.052	−0.354	−0.260	1.000
TANG	−0.084	−0.057	−0.078	−0.047	−0.216	1.000
DB	0.447	−0.080	−0.402	−0.369	0.797	−0.059	1.000
PCT	0.319	0.477	−0.117	−0.076	0.174	−0.101	0.313	1.000
LL	0.176	0.144	−0.214	−0.253	0.671	−0.083	0.641	0.099	1.000
ROA	0.341	0.315	0.177	0.124	−0.242	−0.099	−0.210	−0.057	0.235	1.000

, the correlation matrix is presented, showing no strong correlations (above 0.8).

Table 4. VIF values for the four models.

Model	ROE with C1	ROE with C5	ET with C1	ET with C5
Variable	VIF
RB	4.13	4.22	4.13	4.22
DB	3.56	3.74	3.56	3.74
LL	3.00	3.10	3.00	3.10
ROA	1.68	1.69	1.68	1.69
C1	1.23	-	1.23	-
C5	-	1.19	-	1.19
TANG	1.18	1.17	1.18	1.17
PCT	1.16	1.16	1.16	1.16
Mean	2.28	2.32	2.28	2.32

presents the values for the four regression models, none of which showed a VIF exceeding 5. This led us to conclude that there is no presence of multicollinearity in the sample.

The next step was to apply the Chow test to choose between the pooled model and fixed effects (FE). The results for the four models are shown in

Table 5. Chow test statistics (pooled vs. FE).

	Null Hypothesis: Pooled Model Is the Most Suitable
Model	F-Statistic	p-Value	Reject (R) or Do Not Reject (N/R)
ROE with C1	4.05	0.000	R
ROE with C5	4.11	0.000	R
ET with C1	4.36	0.000	R
ET with C5	4.56	0.000	R

. According to the presented results, for all tested models, the fixed effects model was preferred over the pooled model.

Before performing the Hausman test (to choose between the fixed effects and random effects), it was necessary to identify the presence of heteroscedasticity and autocorrelation; if present, the robust Hausman test must be used.

Table 6. Modified White test for panel data, FE.

	Null Hypothesis: Model Is Homoscedastic
Model	X-Statistic	p-Value	Reject (R) or Do Not Reject (N/R)
ROE with C1	3.10 × 107	0.000	R
ROE with C5	3.10 × 107	0.000	R
ET with C1	6.80 × 1034	0.000	R
ET with C5	3.40 × 107	0.000	R

presents the results of the modified White test for panel data. Based on the results obtained, the null hypothesis of homoscedastic residuals was rejected for all models, demonstrating the presence of heteroscedasticity in the residuals of the generated models.

Table 7. Wooldridge autocorrelation test for panel data.

	Null Hypothesis: Absence of Autocorrelation
Model	F-Statistic	p-Value	Reject (R) or Do Not Reject (N/R)
ROE with C1	2.54	0.121	N/R
ROE with C5	2.84	0.101	N/R
ET with C1	30.19	0.000	R
ET with C5	39.37	0.000	R

shows the results of the Wooldridge test for panel data. The ROE models with C1 and C5 did not show indications of autocorrelation, and with a significance level of 10%, it was not possible to reject the null hypothesis of absence of autocorrelation for these models. In contrast, both ET models with C1 and C5 revealed autocorrelation in the residuals, rejecting the null hypothesis of absence of autocorrelation.

Considering the results from Table 6 (heteroscedasticity tests) and Table 7 (autocorrelation tests), which indicated the presence of heteroscedasticity in all models and autocorrelation in the ET models with C1 and C5, the robust Hausman test was conducted. This test aids in choosing between the fixed effects (FE) and random effects (RE) models. The results are displayed in

Table 8. Robust Hausman test (RE vs. FE).

	Null Hypothesis: RE Model Is the Most Suitable
Model	X-Statistic	p-Value	Reject (R) or Do Not Reject (N/R)
ROE with C1	4.19	0.757	N/R
ROE with C5	5.64	0.583	N/R
ET with C1	1.62	0.978	N/R
ET with C5	2.53	0.925	N/R

. For all proposed models, the random effects model was the most suitable, with a high level of significance (above 50% for all models).

Table 9. Breusch and Pagan test (pooled vs. RE).

	Null Hypothesis: Pooled Model Is Suitable
Model	LM-Statistic	p-Value	Reject (R) or Do Not Reject (N/R)
ROE with C1	27.46	0.000	R
ROE with C5	32.34	0.000	R
ET with C1	68.49	0.000	R
ET with C5	81.12	0.000	R

contains the results of the Breusch and Pagan test for selecting between the pooled model and RE. For all models, the test confirmed that the RE model fits the data best.

Regarding the stationarity tests, the results are presented in

Table 10. Modified ADF test for panel data.

Null Hypothesis: All Panels Contain Unit Roots
Variable	Inverse χ2 (P)	p-Value	Reject (R) or Do Not Reject (N/R)
ROE	779.16	0.000	R
ET	272.22	0.000	R
C1	278.20	0.000	R
C5	161.09	0.000	R
RB	332.62	0.000	R
TANG	282.80	0.000	R
DB	245.88	0.000	R
PCT	672.97	0.000	R
LL	492.00	0.000	R
ROA	539.06	0.000	R

. The tests were conducted without a trend, with deviation, without lags, and with the removal of cross-sectional means. As the results indicate, the series proved to be stationary, corroborating the statistical power of the findings.

In light of the results from the aforementioned tests, the model selected was the random effects model, estimated using FGLS to correct for the heteroscedasticity present in all models and the autocorrelation found in the ET models with C1 and C5. In essence, FGLS is a two-step process that “transforms” the model so that the errors become homoscedastic and uncorrelated; the mathematical explanation goes beyond the intentions of this work. The results of the models are presented in

Table A2. Estimation of the random effects model using FGLS.

	Model
Variable	ROE with C1	ROE with C5	ET with C1	ET with C5
C1	−0.107	-	0.095	-
z-test	−5.540		5.270
p	0.000		0.000
C5	-	−0.020	-	0.103
z-test		−0.630		3.720
p		0.526		0.000
RB	0.035	0.036	−0.109	−0.117
z-test	5.900	5.540	−15.740	−16.740
p	0.000	0.000	0.000	0.000
TANG	0.053	0.052	−0.093	−0.092
z-test	3.810	3.620	−7.340	−7.060
p	0.000	0.000	0.000	0.000
DB	−0.117	−0.105	0.160	0.161
z-test	−16.170	−14.490	28.150	28.180
p	0.000	0.000	0.000	0.000
PCT	0.202	0.195	0.011	0.010
z-test	24.210	23.230	5.360	5.090
p	0.000	0.000	0.000	0.000
LL	0.068	0.060	−0.024	−0.022
z-test	9.580	7.440	−6.700	−6.370
p	0.000	0.000	0.000	0.000
ROA	1.630	1.653	0.556	0.506
z-test	14.530	12.830	8.100	7.580
p	0.000	0.000	0.000	0.000
Constant	0.232	0.110	−0.128	−0.079
z-test	6.090	2.650	−2.650	−1.650
p	0.000	0.008	0.008	0.100

in Appendix B.

In the ROE with C1 model, all variables were significant at a 0% level of significance. Notably, the C1 variable showed a negative relationship with ROE. An increase of 1% in ownership structure results in a decrease of −0.107% in return on equity. The DB variable presented a negative relationship (as expected), while the other variables were positive.

In the ROE with C5 model, all variables were significant at a 0% level of significance except for C5. This variable did not show statistical significance, with a p-value of 52.6%. This indicates that as a company’s capital becomes more dispersed, diversifying control, it is no longer statistically relevant for the performance measured by ROE. The signs and values of the other variables were similar to the ROE with C1 model, reinforcing the correct specification of the model.

For the ET with C1 model, all variables were significant at a 0% level of significance. The C1 variable showed a positive relationship with ET, indicating that as the ownership concentration represented by the control of the largest ordinary shareholder increases by 1%, ET increases by 0.095%, reflecting a deterioration in the company’s indebtedness indicator. RB and LL showed a negative relationship, meaning that as income and revenue increase, ET decreases. The other variables exhibited positive relationships as expected.

In the ET with C5 model, only the constant was not significant at the 10% level, while all other variables were significant at the 0% level. C5 remained significant, indicating that even with capital dispersion, ownership concentration continues to play a relevant role in the company’s indebtedness. This can be explained by the fact that as capital becomes more dispersed, it becomes increasingly difficult to exercise control over indebtedness; decisions regarding debt increases are made in a decentralized manner, showing a slight increase (0.103) compared to the C1 variable (0.095). The other variables showed similar values and signs to the ET with C1 model.

Returning to the research problem presented in the introduction: Does ownership structure influence the debt and efficiency of companies in the electricity sector? The answer to this question was yes. As verified by the results cited above, it can be stated that ownership structure influences the debt and efficiency of companies in the electricity sector. For efficiency, this relationship is negative, but as control is dissipated among a greater number of shareholders, this relationship ceases to exist, demonstrating an improvement in companies that are not concentrated. For debt, the relationship was positive and, despite the dispersion of capital, continues to exert influence. The results found in this study corroborate those found by [11,24,25,26]. The variables used in this study and by the aforementioned authors are similar, if not identical, allowing for comparison. These authors’ analysis focused on the global market, and this research focuses specifically on the electricity sector.

These results indicate potential implications such as a greater likelihood of expropriation of minority shareholders and less professionalization within the company, given the negative relationship found between the concentration of ownership structure and company efficiency. Greater monitoring guarantees may be necessary to prevent expropriation of minority shareholders, even if this increases costs. Improving internal company management, professionalizing key executives, can also contribute to improving these results. Industry regulators can develop policies that encourage competition and capital dissipation, given that greater capital dispersion has improved company performance. These characteristics confirm the results of the authors cited above, stating that the Brazilian market is developing, with room for maturity and efficiency gains with improvements in company ownership structures.

This research has some limitations, including the relatively small sample size and the lack of other sector-specific research for comparison [40]. The results should be interpreted as indicators of the relevance and influence of ownership structure in the sector, encouraging the analysis of these structures as a complementary tool to corroborate company evaluations, but should not be interpreted as determinants of company results or substitutes for traditional indicator analyses.

5. Conclusions

This research aimed to identify the influence of ownership structure on the indebtedness and performance of companies in the electric sector. A panel data methodology was employed to acknowledge the heterogeneity of companies and to measure and identify the actual effects that ownership structure has. This study was innovative, highlighting its originality by measuring these effects specifically for the electric sector, with no similar studies identified for this area until now.

The findings were robust and satisfied various statistical tests, confirming the solidity of the presented numbers. Regarding company efficiency measured by ROE, the results indicated that concentrated ownership structure in the electric sector negatively affects the company’s performance (−0.107). As capital is dispersed, reducing control, the ownership structure ceases to be significant, emphasizing the importance that the largest shareholders exert on the company.

In relation to indebtedness, a positive relationship was found; as concentration increases (0.095), so does indebtedness. Even with control dilution, as measured by C5 (0.103), the ownership structure remains significant. This demonstrates that the ownership structure is relevant in relation to indebtedness; however, even with the dilution of control by the five largest ordinary shareholders, it continues to be significant, necessitating further dilution for the capital structure to cease influencing indebtedness. This could be particularly detrimental to minority shareholders who do not actively participate in decisions regarding company indebtedness.

This work contributes to the scientific literature by confirming that the ownership structure is an indicator to consider in the analysis of companies in the electric sector, as it can affect not only the company’s indebtedness but also its efficiency. This study provides an empirical foundation for a sector that had not been analyzed through this lens, given the relevance and importance of the research theme.

The practical implications of this research’s results lie in highlighting that ownership structure, which is not always considered in analyses, also influences company performance. Regulators can consider ownership structure when developing sectoral policies and use it as an indirect means to achieve their objectives, especially in competition policies and mergers of companies in the sector, considering that the results indicate that more concentrated companies are less efficient. Thus, ownership structure can serve as a vehicle for achieving certain policies. All stakeholders can benefit from these results, as the efficiency of electricity companies can help reduce dependence on fossil fuels.

For future research, studies are suggested that compare the ownership structure of the electricity sector in other countries, allowing the comparison of results from the same sector in different contexts.