Company size: Larger companies have a higher performance and are less likely to be bankrupt and may therefore contract debt more easily (Bhaduri 2002; Titman & Wessels 1988). As a result, it is expected that the company size is positively correlated with debt ratio and company performance.

Company age: The age of a company is measured by the difference between observation year and establishment year. Company age is used as a measure of the reputation in capital structure studies (Abor & Biekpe 2007). Petersen and Rajan (1994) argued that companies with higher debt ratios were more likely to be older, because they were supposed to be of a higher quality. The empirical results indicated that age correlated negatively and significantly with both total debt and short-term debt. Age is also a measure of performance, and the empirical findings indicated that age positively correlated with performance (Glancey 1998; Lumpkin & Dess 2001). This finding is supported by the explanation that older companies may enjoy the benefits of learning and experience, rely on a superior performance and a decrease in efficiency cost, while not being subject to all the liabilities of being new.

Liquidity: The trade-off theory suggests that an increase in liquidity will increase the ability of a company to borrow more, due to a high level of cash or liquid assets to service the interest and principal payment on time (Shahar, Adzis & Baderi 2016). High liquidity ratios allow a company to have a high leverage ratio (debt as a percentage of total capital). This positive relationship is supported by some researchers (Md-Yusuf, Yunus & Supaat 2013; Nadaraja, Zulkafli & Masron 2011). Liquidity is also positively correlated with company performance (Ayaz, Zabri & Ahmad 2021). The positive relationship liquidity and profitability suggests that companies have excess working capital for growth opportunities. This finding is supported by some authors (Deloof 2003; Goddard, Tavakoli & Wilson 2005; Palazoo 2012; Safdar et al. 2016). However, other researchers concluded that liquidity had a statistically significant negative effect on performance (Eljelly 2004; Jose, Lancaster & Stevens 1996; Wang 2002)

Performance: Each theory on capital structure states different relationships to company profitability. Therefore, the ratio between debt and equity will differ (Khan, Bashir & Islam 2020). For example, Khan et al. (2020) argue that companies with positive earnings before taxes, should aim at higher debt ratios to benefit from tax shields as predicted by the trade-off theory. As a result, a positive relationship between leverage and profitability should be expected (Khan et al. 2020). In contrast to this narrative, the information asymmetry and the pecking order theory, suggest that internal funds should be used first for companies with a positive income, because the market will not perceive this as a negative signal suggesting a negative relationship between the variables (Khan et al. 2020). The majority studies in line with the pecking order theory suggested that profitability negatively correlated with debt ratios (Hoque & Pour 2018; Sheikh & Qureshi 2017; Sheikh & Wang 2013).

Leverage According to Ayaz et al. (2021), the relationship between leverage and profitability has been at the centre of attraction for the past decades. Researchers suggested three theories (the trade-off theory, the pecking order theory [POT], and the agency cost theory) which explain the relationship between capital structure and company profitability. These theories take different views on explaining the relationship between the capital structure and company profitability (Ardalan 2017). For instance, the trade-off theory suggested by Kraus and Litzenberger (1973) argues that a company will create a balance between costs and benefits by using debt financing. As a result, managers are interested in increasing the use of leverage, which eventually maximises company performance, suggesting a positive relationship (Ayaz et al. 2021; Zeitun & Saleh 2015). On the other hand, the POT proposed by Myers and Majluf (1984) indicates that a company prioritises its funding sources from internal (retained earnings and depreciation) to external (debt) and to equity, suggesting an inverse relationship between debt ratios and performance.

To investigate the effect of BO on performance and capital structure, data were sourced from the Iress database, a prominent source of financial data in South Africa, and annual reports for the period 2007–2014. This period is characterised by the 2008 financial crisis. One hundred and eighty-seven JSE-listed companies, without any significant gaps over the sampled period, were selected from eight sectors (the basic materials, consumer goods, consumer services, healthcare, technology, telecommunication, industrials, as well as oil and gas). Therefore, the data consisted of a panel of 187 companies observed over an 8-year period providing a total of 1496 observations. Winsorisation was applied to the data at the first and 99th percentiles to minimise the problem of outliers in both cross-sectional and panel regressions.

The study was a deductive model based on an in-depth analysis of cross-sectional and time-series pooled data from the audited annual reports of JSE-listed companies. It was also a causal study because it sought to demonstrate the effect of BO on performance and capital structure. To investigate the relationship, the research used fixed-effects and random-effects models. According to Vieira, Neves and Dias (2019) panel data models have a number of benefits over other models. These advantages include greater variability of data, more information, more degrees of freedom, good control of the impact of omitted company’s, more accurate inference and greater control of the endogeneity (Baltagi, Bratberg & Holmås 2005; Hsiao 2007). However, according to some authors (Ayaz et al. 2021; Gaud et al. 2005; Le & Phan 2017), the fixed-effects model may generate estimates that are biased and inconsistent because of the endogeneity. Some authors (Ayaz et al. 2021; Ullah, Akhtar & Zaefarian 2018) argue that not only will the results be biased and inconsistent, but they will also yield misleading conclusion and unsuitable theoretical understanding. As a result, in the research a dynamic two-step generalised method of moments (GMM) estimation technique was used to alleviate the endogeneity problem and produce consistent estimates (Bänziger 2018). To determine the empirical impact of BO on capital structure and performance following prior research (Ayaz et al. 2021), the following regression models are used:

The variables are defined in Table 1.

In this section, the results of the study are presented and discussed according to the literature review and the formulated hypotheses. Table 2 provides a summary of descriptive statistics including minimum, maximum, mean, standard deviation, the 50th percentile and the 99th percentile values of all variables. The three measures of the capital structure (Short-term debt, long-term debt and total debt based on market value) recorded overall means of 26%. (0.255), 14% (0.143) and 36% (0.358) respectively and dispersed from 0% to 91%, 0% to 65% and 0% to 100%. The finding suggests that most companies in the sample were more equity financed than debt financed. In addition, the finding suggests that South African companies listed on the JSE used 36% of debt and were less indebted than companies in China (Zou & Xiao 2006), Germany (Abdullah & Tursoy 2021) and Vietnam (Le & Phan 2017). However, the debt ratio was higher than companies in France (Lin et al. 2013). Tobin’s Q (TQ) was 148% for the full sample. Return on assets (ROA) and return on equity (ROE) recorded overall means of 10% and 12%. The performance measures indicated that South African companies performed poorly during the period 2007–2014. Some of the reasons for the low ratios may be the 2008 financial crisis and low-rate participation of the proportion of BO. Further, the market gap in performance was highlighted by maximum and the minimum. The average BO proportion out of the total ownership structure was 0.3% (029), suggesting that direct BO represented about 3% of the total ownership. The average company size in the sample was 6.239, calculated as the logarithm of total assets. The average age of JSE- listed companies in the sample was 25.037 years. The liquidity registered an overall mean of 2.189. The variable that presented the highest standard deviation is age.

To check for multicollinearity problems, a correlation analysis was used. Table 3 reports these results. The results in Table 3 revealed that the coefficients were less than the value of 0.80, indicating that the variables were not highly correlated (Ayaz et al. 2021). The correlation results show that debt ratios (short-term debt and total debt based), company size and ROE had significant and inverse relationships with direct BO. Performance measures inversely and significantly correlated with debt ratios. Company age inversely and significantly correlated with the total debt ratio. Liquidity negatively and significantly correlated with debt ratio. Company size positively correlated with leverage and performance. This finding supports the notion that large companies have more assets for collateral and it is easier for them to negotiate better terms with lenders (Butt & Hasan 2009).

To determine the effects of direct BO on debt ratios and performance, a fixed-effects model, a random-effects model and a two-step generalised method of moments were used. Table 4 presents the results of the fixed-effects model (Equation 1–3) and the random-effects model (Equation 4–6) for the effect of BO on capital structure; Table 5 presents the results of the fixed-effects model (Equation 1–3) and the random-effects model (Equation 4–6) for the effect of BO on performance. The fixed effect model is the best model for the interpretation and discussion based on the Hausman (1978) test. The results of the two-step generalised method of moments are presented in Tables 6 and 7 for the capital structure and performance, respectively, and their relationship with BO. Many of the research results were virtually the same when using the different measures of capital structure and performance suggesting that the findings are robust and valid. Furthermore, most of the coefficients estimated, had the expected signs and were significant. The results showed that BO was not significant in both the capital structure and performance equations in the fixed-effects and random-effects models. However, in the generalised method of moments, the coefficient of BO inversely and significantly correlated with long-term debt and TQ. As a result, hypotheses 1 and 2 are rejected. This finding supports the prediction that companies with a relatively small proportion of BO cannot support high leverage and high performance. However, the negative relationship might have been caused by other reasons. For example, debt may be restricted to BO. Surprisingly, the coefficient of BO positively and significantly correlated with ROA. The coefficient of the dummy variable, BO and total ownership, indicated that the proportion of leverage was lower for BO than for total ownership. The dummy variable was insignificant in the performance equations in the fixed- and random-effects models but significant in the generalised method of moments. This finding indicated that performance was higher for BO than for total ownership when the dependent variable was the TQ. However, performance was lower for percentage of BO, than for total ownership, when ROAs was used as the dependent variable. Profitability had a negative and significant relationship with all debt ratios, while long-term debt based on market value (LLM) negatively correlated with all performance. The negative relationship suggests that South African companies prioritise their funding sources from internal to external, and debt to equity (Ayaz et al. 2021). There was strong evidence that company size had a positive effect on leverage ratios and performance measures. The positive relationship may suggest that large companies may be more diversified and fail less often. Age positively and significantly correlated with debt ratios and negatively and significantly correlated with performance. The positive relationship between age and leverage is in line with the narrative that older companies have higher debt ratios, because they feel somewhat obliged to present themselves as higher-quality companies, while the negative correlation between age and performance contradicts the notion that older companies are more experienced, enjoy the benefits of learning, are not prone to the liabilities of new companies and therefore enjoy superior performance and a decrease in costs.

In the generalised method of moments, the empirical findings indicated that the coefficients of the lagged three measures of capital structure were positive, as expected. The adjustment coefficient of LLM was relatively large (more than 0.5), possibly providing evidence that companies adjusted their debt ratio ratios faster to reach their target debt ratios.

The cost of deviating from target debt ratios is a possible explanation for this adjustment speed. However, for short-term and total debt, the adjustment coefficient was smaller. The lagged coefficients of performance were positive and insignificant. The finding indicated that past realisation of performance was not significant in explaining future realisation of performance. The coefficient of company profitability inversely and significantly correlated with debt ratios. The findings also indicated that industry effects were significant in explaining financing decisions and performance in some sectors relative to other sectors.