Institutional investors and dividend payments: evidence in the oil industry

Abstract

The debate over the impacts of shareholder value orientation on corporate management has been more intense with the increasing participation of institutional investors in companies’ ownership structures. In this context, the purpose of this study is to evaluate the influence of institutional investors’ shareholding on the payment of dividends in the oil industry. A regression model was used, estimated with the Generalized Method of Moments. The results indicated that the distribution of dividends is related to the profitability and the leverage of the companies, in addition to the history of distribution to shareholders. In general, the presence of institutional investors did not influence the dividend distribution. However, we observed a large participation of these investors in the ownership structure of companies in the oil and gas sector—the average control of these agents was around 25% in the companies of the sample. This study contributes to the literature regarding the influence of institutional investors on the corporate decisions of nonfinancial companies, being original in the context of the oil industry.

1. Introduction

Oil is the main source of primary energy in the world economy. The dependence on this commodity is presented in multiple forms throughout the daily life of modern society, in a long chain of derivatives. This industry combines operational activities of production and exploration with implicit power relations. As it gained prominence in the world’s industrial economies, its circulation became vital to ensuring the stability of a country’s energy security.

Throughout the history of the oil and gas industry (O&G), companies with significant financial structures have been formed, being among the largest and most profitable corporations in the world. With the changes in the energy, industrial, and financial paradigm, O&G companies have been going through a transition period, which materializes in a strategic redirection of these corporations to maintain the level of investment in their main activity, direct investments for energy transition, and ensure adequate payment of dividends to their shareholders (Pickl, 2019, 2021).

Focusing on dividend policy, recent studies have investigated whether the phenomena of financialization of nonfinancial companies has been influencing dividend payments (Driver et al., 2020; Yu and Jo, 2022; Valeeva et al., 2023). Financialization is a new regime of accumulation, in which the productive force has been progressively controlled by the financial markets—in other words, a finance-dominated accumulation regime. The “retain-and-reinvest” based allocation criterion has been replaced by “downsize-and-distribute” as transfers in the form of interest, dividends, and share buyback programs (Davis, 2009; Rabinovich, 2019). This process has influenced the management and corporate performance of these companies, which are embodied in the concept of maximizing shareholder value (MSV). This principle predominated as an instrument of corporate governance that increases the importance of shareholders over the stakeholders (Lazonick and O’Sullivan, 2000).

The shareholder-value orientation assumed prominent importance from the moment that institutional investors became a relevant part of the financing of nonfinancial companies, assuming important positions in the ownership structure of these corporations. In this context, recent studies have analyzed the influence of institutional investors on the most different strategies adopted by nonfinancial companies (Dyck et al., 2019; Hasan et al., 2021; Li and Ji, 2021; Kristanti and Ardiningrum, 2022; Nurokhmah et al., 2022; Hazmi et al., 2023). Generally, results have indicated that such influence varies according to the characteristics of these investors (Crane et al., 2016; Karwowski and Stockhammer, 2017; Fonseca et al., 2019; Al-Qadasi et al., 2022; Ataullah et al., 2022). Most of the literature analyzes institutional investment in a set of nonfinancial companies without considering sectoral stratification. Thus, given the importance and magnitude of the oil sector in the world, the analysis of how institutional investors influence the distribution of dividends and the corporate strategy of companies in this industry is an important gap to be filled.

The objective of this study is to investigate whether the presence of institutional investors in the ownership structure of oil and gas companies influences the payment of dividends to shareholders. We hypothesize that the participation of institutional investors influences the dividends distribution. The O&G sector is widely understood to have reached its mature phase, with repercussions on the composition of companies’ cash flow and attenuation of the growth rate of capital spending. These are companies with large capacities to generate cash. Thus, the difference between the reduction of capital expenditures and the maintenance of cash generation capacity may be redirected to the payment of dividends and the repurchase of shares. On the other hand, with the change of the energy paradigm, due to the pressure to reduce environmental damage, this industry has been going through a period of transition to adapt to the forces that are leading to a low-carbon economy; which may be also reorienting investments and redirecting the strategies of these corporations and their investors.

This study contributes to the literature regarding the influence of institutional investors on the corporate decisions of nonfinancial companies, being original in the context of the oil industry. The evidence can provide interesting insights to industry stakeholders and executives, especially in a context in which companies are adapting to a transitional business model. Moreover, the results can contribute to academic discussions about the financialization of nonfinancial companies and to the debate about the influence of shareholders on corporate strategies, highlighting the importance of evaluating such aspects in a sector of broad importance and dynamism.

2. Literature review

The process of financialization is connected to the expressive growth of a class of agents called Institutional Investors. This group is formed by pension funds, hedge funds, investment funds, and insurance companies, among others. Defined as financial intermediaries specialized in financial investments, these agents can act independently or be part of a group or conglomerate. In essence, they manage third-party capital and have become the main managers of collective savings (Farnetti, 1998; Plihon, 2005; Sauviat, 2005; Carneiro, 2010; Çelik and Isaksson, 2013; OECD, 2014).

This set of investors has exerted an important influence on the capitalist dynamics dominated by the financial regime (Çelik and Isaksson, 2013; OECD, 2014). In general, these institutions have the potential to discipline and define standards and guidelines for companies with the legitimacy achieved by MSV.

In addition to the penetration into the ownership structure of nonfinancial corporations, institutional investors assumed an important position in the centralization of debts and funding the activities. Since the 1980s, large corporations have issued bonds for capital financing. Concomitantly, the management of financial reserves has been transferred from traditional deposit institutions, such as commercial banks and savings institutions, to institutional investors. This transition perpetuated the participation in the liabilities of the companies by funds and insurers. Additionally, managers, who were under pressure to generate results, have applied a large part of these resources to leverage the profitability of portfolios. The growing performance of institutional investors has increased these agents’ ability to influence the behavior of companies to maximize shareholder value. Thus, the directional and normative power of these investors derives from the dual condition that these companies have undertaken within the corporations. On the one hand, as shareholders, and on the other, as their main lenders. Thus, their power unfolds as both the owners and creditors of large corporations.

The influence of institutional investors on corporate governance has been the subject of debate in the literature. Since these agents form a heterogeneous group, their effect on corporate management tends to diverge (Froud et al., 2000; Fonseca et al., 2019; Lazonick and Shin, 2020). The asymmetry of the impacts caused by these investors arises from the fact that the level of involvement in management is determined by the decision-making of each institution, according to its “business model” (Çelik and Isaksson, 2013) and its monitoring efforts of the targeted corporation (Katan and Mat Nor, 2015).

Fonseca et al. (2019) separate the authors who evaluated the effects of the participation of these investors into two main groups. On the one hand, the presence of institutional investors in the ownership structure may be correlated with proactive effects, seeking to ensure the sustainability of the company and its growth. On the other hand, a group of authors reports a sort of “shortsightedness” of these agents, who push for short-term gains.

Crane et al. (2016), Bataineh and Ntim (2021), and Hazmi et al. (2023) argue that greater participation by institutional investors causes higher dividend payments. For Nurokhmah et al. (2022), these investors have the potential to increase the payout ratio of companies. However, we can also find recent studies that do not corroborate this evidence. Fonseca et al. (2019), for example, point out that these investors do not influence dividend policy. On the other hand, they have a positive influence over financial investments. Similarly, Hasan et al. (2021) and Kristanti and Ardiningrum (2022) found a significant and negative effect between the participation of these investors and the payout ratio.

Ferreira and Matos (2008) contribute to this debate, showing that institutional investors have a strong preference for shares of large companies and companies with good governance. The presence of independent and foreign institutions in the ownership structure of companies strengthens the shareholder value and the operational performance of corporations. Chen et al. (2007) highlight that long-term-oriented investors seek to monitor operational activities rather than seek short-term gains. Ataullah et al. (2022) add that the heterogeneity of institutional investors is a potential factor in corporate decisions. These authors observed that the greater the control of these investors, the lower the sensitivity to paying dividends in times of external shocks, such as the pandemic.

Based on such analyses, we observed that evidence does not follow the same pathway, varying according to the country and period analyzed, and the specificity of the companies. This study evaluates this topic by focusing on the O&G industry, seeking to contribute to the debate by analyzing an important sector of the world economy.

3. Methods

3.1 Econometric model

To investigate the evidence of financialization in the oil industry, we use a model that explores the relationship between dividend payments and the participation of institutional investors in the ownership structure of O&G companies. We assumed the existence of a reciprocal relationship between companies and financial markets—while the higher the potential for financial return in the form of dividends, the greater the incentives for the entry of these investors (Jain, 2007; Nguyen and Li, 2020) on the other hand, this presence tends to push for a greater distribution of the net profit (Crane et al., 2016; Bataineh and Ntim, 2021).

The problem with the interdependent relationship and simultaneous determination between variables, quite common in corporate finance literature (Barros et al., 2020), can be mitigated by using the Difference Generalized Method of Moments (GMM-Dif), developed by Arellano and Bond (1991). Equation (1) describes the model:

In which, i and t represent, respectively, the i-th cross-sectional unit (companies) and t-th the moment of time (years). The dependent variable, Div_it, represents the ratio between the dividends distributed by companies and total assets. The intercept is represented by |${\beta _0}$|⁠; |$\beta _1^t{\rm{\,and\,}}\beta _2^t{\rm{\,}}$|represent the transposed vectors of parameters. In addition, |${x_{it}}$| is the vector of explanatory variables, which is divided into two subvectors: endogenous or predetermined (⁠|${x_{1it}}$|⁠) and exogenous (⁠|${x_{2it}}$|⁠). Finally, |${a_i}$| and |$\,{\mu _{it}}\,$|represent, respectively, the random effects and the error of the model term, considering that |$\mathbb{E}\left[ {{\mu _{it}}} \right] = {\rm{ }}\mathbb{E}\left[ {{\alpha _i}} \right] = 0$|⁠.

In the model developed by Arellano and Bond (1991), the estimators are based on the difference between the variable to be explained and the endogenous variables, and also on the difference between the strictly exogenous variables. In this case, first, the variables are transformed into the “first difference” to eliminate the unobserved heterogeneity. Here the procedure consists of calculating the differences of the variables about their lagged values. Then, we can accommodate the hypothesis that endogenous regressors are correlated with past values of the error term, but not with their present or future values, configuring a sequential exogeneity of the regressors (Fonseca et al., 2019; Barros et al., 2020; Rabinovich and Pérez Artica, 2022).

However, as established in the studies of Arellano and Bover (1995) and Blundell and Bond (1998), the model estimated by GMM-Dif, although capable of conceiving asymptotically valid and consistent statistics, can generate inaccurate and biased estimates. To deal with these issues, Blundell and Bond (1998) elaborated an extension of the method, called the System Generalized Method of Moments (GMM-Sys), used in this study. In this approach, the estimators are combined in a system of regressions in differences with regressions in level, using the lags of the endogenous explanatory variables in differences as instruments. If the sequential exogeneity hypothesis is validated, other momentum conditions are imposed on GMM-Sys, equations (2) and (3):

Therefore, an additional premise is assumed: |${{\Delta }}{x_{1it - 1}}$| and |${a_i}$| are not correlated, even if the correlation between the regressors (explanatory variables) and the unobserved heterogeneity is feasible. This premise holds as long as the form of the correlation does not change over time. Therefore, in GMM-Sys, the same momentum conditions as GMM-Dif are used, and others are added, increasing the efficiency and performance of the estimator (Blundell and Bond, 2000; Barros et al., 2020).

The modeling of GMM-Sys occurs in two stages. First, it is necessary to assume the premise that the errors are independent and homoscedastic between the cross-sectional units and along the period. Second, the residues that were obtained in the previous stage are used to determine a consistent estimate of the variance–covariance matrix, allowing for flexibilization of the hypotheses of independence and homoscedasticity. However, the two-stage estimator can generate underestimated standard errors. To deal with this problem, Windjmeijer correction for finite samples was used, ensuring that two-stage results are more efficient and standard errors are not biased (Roodman, 2009; Fonseca et al., 2019).

Data availability may vary within a time interval. In this case, the first difference transformation tends to maximize empty spaces in unbalanced panels. A |${y_{it}}$| missing will have implications on |$\Delta {y_{it}}$| and |$\Delta {y_{it,\,t + 1}}$|⁠. This weakness motivated the use of orthogonal deviations, proposed by Arellano and Bover (1995). In this approach, instead of subtracting the previous observation from the current one, the average of all available future observations is used. This minimizes data loss and keeps outdated observations as valid instruments. Additionally, corrections were used for small samples (small command) in the estimation of the covariance matrix, resulting in t-test statistics, to the detriment of the z-test for the coefficients, and an F-statistic instead of the |${\chi ^2}$| Wald statistics to verify the overall fit of the model (Roodman, 2009).

To validate the statistics and ensure the estimation hypothesis of the model (absence of correlation between endogenous instruments and regressors, but not between instruments and the error terms) the following tests were applied: (i) first- and second-order autocorrelation test—AR(1) and AR(2), developed by Arellano and Bond (1991); (ii) Hansen test of overidentifying restrictions.

The tests proposed by Arellano and Bond (1991) seek to evaluate the hypothesis of autocorrelation in residuals in difference. The null hypothesis to be tested is that there is no autocorrelation, and the statistic will have autocorrelation in the first order in the idiosyncratic errors in the first difference, but not in the second. Therefore, we expected the null hypothesis to be rejected by AR(1), obtaining a negative statistic, and that the statistic is not significant in AR(2).

The Hansen test qualifies the instruments included in the model (overidentifying restrictions or additional timing conditions). The null reference hypothesis for the test is that all instrumental variables are uncorrelated with the error term. Therefore, we expect the null hypothesis not to be rejected, which means signaling that some of the variables used are not exogenous, validating the employed moment conditions.

3.2 Data and variables

The sample was based on the 50 largest companies in the oil and gas sector,¹ using the 2010–2020 period. Thus, we have panel data, structured from cross-sectional units applied in time series. Tracking preestablished units (O&G companies) over a certain period provides advantages—more informative data analysis—allowing us to absorb elements of individual heterogeneity and the effect of the lag over time (Baltagi, 2005; Wooldridge, 2006).

The dependent variable of the model is the ratio between the amount distributed to shareholders as dividends and the total assets of the company i in year t—equation (4).

To verify the influence of institutional investors on Div_it, we considered the following control variables:

• Size (SIZE): the natural logarithm of sales was used as a size proxy (Titman and Wessels, 1988). The O&G sector is composed of large companies, mostly with an advanced degree of maturity. DeAngelo et al. (2006) pointed out that the companies that distribute the most dividends are mature. Fama and French (2001) suggest that companies with lower growth rates are more likely to pay more dividends. Ferreira and Matos (2008) pointed out that institutional investors have a strong preference for shares of large companies. Therefore, a positive estimated coefficient is expected.

• Capital Expenditures (CAPEX): fixed assets expenses represent the largest component of oil companies’ cash flow, and they compete with the payment of dividends. Thus, a negative sign is expected for the relationship between CAPEX and the dependent variable (Weijermars et al., 2014).

• Profitability (ROE): the metric used to evaluate the profitability of companies was Return on Equity. Companies with higher profitability are likely to pay more dividends (Fama and French, 2001; Brav et al., 2005; DeAngelo et al., 2006). Thus, a positive estimated coefficient is expected.

• Leverage (LEV): the degree of financial leverage was analyzed through the relationship between total debt and equity. On one hand, higher leverage can cause a restriction on future cash flow, due to amortization and interest. On the other hand, higher leverage can increase investments and induce a cash expansion, which may result in more dividends a posteriori (Jensen et al., 1992; Kliman and Williams, 2015; Fonseca et al., 2019). Therefore, the relation between this control variable and dividend payments is undefined.

• Current ratio (CR): indicates the financial flexibility of the company. Companies that are more liquid or have more cash capacity have a greater potential to distribute dividends (Acharya and Viswanathan, 2011; Fonseca et al., 2019). The expected estimated coefficient is positive.

• Lag of the ratio between dividend payment and total assets |$(Div/A{t_{it - 1}}$| and |$Div/A{t_{it - 2}}$|⁠): the dividend projection is influenced by previously paid dividends, given that agents expect a positive progression of distributed earnings (Lintner, 1956). O&G companies are mature companies with a good track record in paying dividends (Labban, 2014). Therefore, the hypothesis is that the dividend policy is maintained, and the estimated coefficient sign is positive.

In addition, we also used dummies for years in the model to isolate the effects of shocks not related to the estimated model, capturing the fixed effects of time, such as changes in the international price of oil, and macroeconomic cycles, among other variations in the economic situation that may alter the allocation of resources by companies. These dummies help to maintain the hypothesis that the estimates of standard errors are not correlated with individuals (Roodman, 2009).

Two different model specifications were estimated, considering two possibilities to assess the impact of institutional investors on the payment of dividends. In the first, variable II₁ was used, which was obtained by the ratio between the number of ordinary shares (ORD) held by institutional investors and the total of ORD shares.²

In the second specification, we used a dummy variable, II2, that assumes a value equal to 1 when the company has II1 variable in the top quartile group (top 25% of the distribution); and 0 for the others. In this group, the participation of II in the companies’ ownership structure is higher than 45%. The use of this variable aims to identify if the institutional investors with a notable participation in the companies’ ownership structure have a special influence on dividend policy. The sign of the coefficient of these variables is expected to be positive.

Table 1 summarizes the variables of the model, their definition, and the expected sign of the relation between the dependent and explanatory variables. Among the variables used, past dividends and the presence of institutional investors (II₁ and II₂) were defined as endogenous, and the remaining variables as exogenous.

3.3 Descriptive analysis of data

Table 2 presents the descriptive statistics of the sample and Table 3 shows the correlation matrix. The ratio between dividend payments and total assets (DIV) was, on average, 1.8% per year, with a maximum of 14.8%. The ROE of the companies presented a mean of 8.42%, with a high standard deviation (15.13%). Leverage (LEV) showed a mean level of 63.61%, also presenting a high standard deviation (60.12%). The mean natural logarithm was 24.35 for sales (SIZE) and 7.21 for capital expenditures (CAPEX).

The average control of institutional investors was around 25% for the companies in the sample, ranging from zero to 83%. The literature shows that the control of these investors in the ownership structure varies according to the sample of companies and the definition used for the classification of institutional investors. Al-Sartawi and Sanad (2019) evaluated this share in a sample of 48 nonfinancial companies listed in Bahrain and found an average share of 51%. Musallam et al. (2019) reached an average share of 49% between 2009 and 2013, while Kristanti and Ardiningrum (2022) registered 61% share between 2013 and 2020, both in Indonesia. Panda (2023) recorded an average share of 21% for a sample of nonfinancial companies from India.

Regarding the correlation coefficients, it is verified that the profitability and the current ratio presented a positive and significant correlation with the payment of dividends. On the other hand, leverage had a significantly negative correlation.

Figure 1 shows the distribution and descriptive statistics in each year of the dependent and explanatory variables. Each graph shows the median (point) and the interquartile range (vertical bar in the center). It is not possible to identify a distribution with a similar format to normal between the variables used. Notably, the variables SIZE, CAPEX, and II₁ presented no structural changes between the 2010 and 2020 period. On the other hand, DIV, ROE, LEV, and CR showed significant variation.

Figure 2 illustrates the presence of institutional investors in O&G companies’ ownership structure in 2020.³ The dots represent the companies (black dots) and investors (white dots with gray lines), and the arrows, their investments. The greater the number of institutional investors as shareholders, the greater the label with the name and the node. The thickness of the arrow indicates the size of the shareholders’ investment. The network considers only II investments above $1 billion to help identify the most important and central players. It is possible to infer a high presence of this class of investors in the ownership structure of the sector. Notably, the largest investments in the US majors were received from Vanguard and State Street, with a particular focus on Exxon Mobil, Chevron, and ConocoPhillips.

4. Results

Table 4 shows the results of the estimation of the two models, in the first and second stages. The estimated coefficients, standard deviations (in parentheses), and significance level are included. For all coefficients, the correction of standard errors proposed by Windjmeier for finite samples, orthogonal deviations, and corrections for small samples was employed.

The results indicate a robust fit of the model to the data. In Hansen’s test of overidentifying restrictions, in all models and specifications, it was not possible to reject the null hypothesis, attesting to the suitability of the instruments used in the proposed model, and signaling that the instruments are not correlated with the error terms. In turn, the autocorrelation tests developed by Arellano and Bond (1991) found the presence of first-order negative autocorrelation in the idiosyncratic errors of the first difference, rejecting the null hypothesis, but it was not possible to reject the second difference, given that the statistic was not significant. Therefore, it is possible to corroborate the hypothesis that |${\mu _{it}}$| it is not autocorrelated in all the variables of the four models.

Since the model is statistically robust, the interpretation of the coefficients is valid. The signs of the coefficients in the first and second stages are identical, and the values of the coefficients are not significantly altered, which attests once again to the validity of the models. The two-stage estimator is more efficient, albeit biased. However, with the Windmeijer (2005) correction, the two-stage specification becomes asymptotically more efficient and unbiased (Roodman, 2009; Rabinovich and Pérez Artica, 2022). Therefore, the interpretation of the model will be concentrated on this specification.

The results indicate that the dependent lagged variable in one and two periods is statistically significant and positively correlated with its current value in all specifications, which ratifies the hypothesis of a stable dividend policy for the companies in the sample. Return on equity (ROE) had a positive and statistically significant effect on both models. In addition, financial leverage (LEV) presented a negative and significant sign. Therefore, the restriction on future cash flow acts as an adverse element to the distribution of dividends, with effects that overlap cash availability. On the other hand, size, CAPEX, and the current ratio did not present significant statistics.

Regarding the ownership structure, both in the model that evaluates the participation in common shares (II₁), and in the specification that evaluates the high presence of this class of investors (II₂), the coefficients were not significant to explain the distribution of dividends in the sample of O&G companies.

The literature not only supports the results obtained by the econometric analysis, but it also raises new questions for the O&G sector. Firstly, the positive and significant representation of the dependent lagged variable lagged in two periods ratifies the assertion that the sector is permeated by companies with a good track record in paying dividends (Labban, 2014) and, therefore, the payment of past dividends influences the current distribution. This movement is due to pressure on the board and top management to meet investors’ expectations.

Another aspect that has already been observed in the literature is linked to the relationship between the company’s profitability and shareholder return. Previous studies have reported the importance of this variable in investment and dividend policies. Gill et al. (2010) evaluated the US service and manufacturing companies, and the results indicated that dividend payments are a function of profit margin and sales growth. Mehta (2012) analyzed companies listed on the Abu Dhabi Stock Exchange from 2005 to 2009, showing evidence that profitability and size are the most important elements in dividend decisions. According to DeAngelo et al. (2006), the higher the company’s profit, the greater the pressure to distribute part of the earnings to shareholders. In the O&G sector, Weijermars et al. (2014) indicated that the highest profitability achieved in some years by companies in the sector has been converted into MSV as dividends. Therefore, the significant coefficient for return on equity contributes to the ratification of the hypothesis.

Another important result was the negative and significant coefficient for financial leverage. According to Macchia La et al. (2017), the greater the participation of third-party capital in the financing structure, the lower the degree of freedom over financial decisions. Fonseca et al. (2019) concluded that companies with higher financial leverage pay fewer dividends. Increasing debt in the present means restriction of future cash flow, corroborating with the negative sign of the coefficient.

Regarding the ownership structure and the importance of institutional investors in the distribution of dividends, the evidence presented in the literature is not conclusive. The results showed that, in all specifications, the participation of institutional investments in the control of O&G companies is not significantly related to the payment of dividends, in line with recent studies—Fonseca et al. (2019) and Hasan et al. (2021), for example. This result is justified by a set of conjugated factors. First, it should be noted that these investors form a heterogeneous group, with distinct characteristics, strategies, time horizons, and business interests (Froud et al., 2000; Fonseca et al., 2019; Lazonick and Shin, 2020; Ataullah et al., 2022). This diversity hinders an integrated action capable of acting effectively and in a targeted way on the allocation of resources.

Another issue raised by Çelik and Isaksson (2013) relates to the asymmetry of effect, which, in turn, varies according to the involvement of each institution with the asset. The business model materializes a multifaceted set of interests. Therefore, the decision on dividends competes, for example, with investment decisions, both in new oil reserves and in investments in renewable energies. The maintenance of these companies as important players in the global productive dynamics will depend on the transition from the original business model to a more wide-ranging model, in which companies diversify the lines of business under the prism of energy companies and incorporate the enabling technologies of the energy transition (Pickl, 2021). Another important element is related to the maintenance period of the asset in the portfolio and the monitoring efforts of the targeted corporation (Katan and Mat Nor, 2015). The shorter the time, the less chance of monitoring and pressuring the management of companies. Finally, not all investors put themselves in a short-term vision for return, part can support the reinvestment of profits under a longer-term view (Chen et al., 2007).

5. Conclusions

The objective of the research was to evaluate the main factors that influence the payment of dividends in the O&G industry, especially the participation of institutional investors. The results indicated that the distribution of dividends is related to the history of distribution to shareholders, profitability, and financial leverage, but not to the participation of institutional investors. Another important contribution is the identification of the strong presence of institutional investors in the ownership structure of companies in the oil sector, especially in a sample that incorporates state-owned companies. The evidence showed that, if there is a minimum number of outstanding shares, these managing entities will compete for voting shares.

These results are supported by the literature and add important and original elements to the understanding of the O&G. Notably, the history of dividend distribution, profitability, and leverage are preponderant factors in the definition of dividend policy. Concerning the ownership structure, the explanation for the nonsignificant coefficients for the presence of institutional investors lies in the level of heterogeneity of this class of investors with distinct interests, in addition to the recent transformations that affect the sector. The energy transition is imposed by international agreements to reduce carbon emissions and by social pressure to increase the share of renewable energy, with effects on the business model and resource allocation of large oil corporations.

This study provides new insights into the determining factors of dividend policy in the O&G industry. It also leaves more research questions to be investigated in future research. This topic can be further explored by deepening the analysis of the heterogeneity of institutional investors that participate in the ownership of O&G companies. This point can be particularly relevant to understanding whether the lack of interest of certain types of institutional investors in interfering in management decisions is linked to short-term factors, such as portfolio diversification. In addition, studies can explore if the maintenance period of the investment of the institutional investors influences the dividend policy; and analyze whether indirect participation—via affiliates and subsidiaries—, and the level of complexity of the network of action on the ownership structure affects the strategic direction of the companies. These are some central questions that can contribute to the understanding of a historically important industry that has been adjusting to the new technological, environmental, and financial paradigms.

Acknowledgements

The authors thank CAPES (Coordination for the Improvement of Higher Education Personnel – Brazil) for the financial support of this research.

Conflicts of interest

All authors declare that they have no known competing financial interests or personal relationships (directly or indirectly) that could have appeared to influence the work reported in this paper.

Footnotes

References