Static trade-off theory

The theory presents the idea that there is an optimal capital structure. There is a combination of equity and debt that is able to increase the value of the company. Debt utilization generates two opposite implications: tax benefit (positive effect) and bankruptcy costs (negative effect) (^{Santos et al., 2009}). It is necessary for companies to seek a capital structure that will maximize the benefits and minimize debt costs (^{Bastos; Nakamura, 2009}).

According to ^{Myers (1984)}, companies should gradually replace equity with debt and vice versa to an equilibrium point at which it is possible to reduce the costs and increase the value of the company.

Asymmetric information and Pecking Order Theory

Pecking Order theory is the approach that determines the optimal capital structure. It is based on minimizing inefficiency in the companies’ investment decisions (^{Myers; Majluf, 1984}; ^{Myers, 1984}).

The idea presented by ^{Myers and Majluf (1984)} is based on information asymmetry between managers and investors, i.e., managers have more and better information on investment options and financing, to the point of prioritizing the opportunity that will bring lower risk to the company.

Thus, ^{Myers (1984)} developed the Pecking Order Theory that determines a hierarchy in the sources of funding: firstly, using the retained earnings as sources of financing; secondly, acquiring new debt; and, ultimately, issuing shares to obtain resources. This structure is maintained even if the shares are overvalued.

Market timing theory

In a globalized environment it is clear that a range of factors can influence the value and the capital structure of an organization, and according to ^{Frank and Goyal (2009}) managers follow two lines regarding capital structure decisions: suspend the financing decision for as long as possible if the conditions are not favorable or take advantage of unusual opportunities even if there are no needs.

The Market Timing Theory explains that a company with greater financial leverage obtains funds when its market value is low. Companies that show lower leverage capture external resources when the market value is high (^{Baker; Würgler, 2002}). ^{Hovakimiam (2006)} stated that in a warm market the issuance of shares that are held exerts no influences on capital structure. In contrast, ^{Alti (2006)} shows that in a warm market companies enjoy the opportunity to issue more shares, overvaluing them and reducing debt. However, after overvaluation period companies increase their debts.

However, ^{Alti (2006)} also shows that the theory is an important determinant of financing activity, but only for the short term.

Agency theory

Debt utilization is key to the emergence of conflicts of interest between shareholders and managers (^{Ross; Westerfield; Jaffe, 2002}). Thus, because of such conflicts, ^{Mendes et al. (2009)} show that the agency theory can be understood as the costs that come from the purpose of resolution of these conflicts, since managers and shareholders intervene in defense of their own interests.

^{Grinblatt and Titmman (2005)} elucidate that high company debt may be more favorable for shareholders as it would limit managers to adopt effective measures in personal terms and to expand the company more rapidly than advisable. On the other hand, managers prefer a smaller debt than the optimal point because it reduces the bankruptcy risk and also does not restrict management.

According to this theory, the optimal debt is defined by minimizing the total agency cost (^{Mendes et al., 2009}).

The economic crisis variable

Effects of the globalization of economic crises have many critical consequences for public companies (^{Berger; Turtle 2011}). According to ^{Espinola (2013)} the cause of crisis is related to financial bubbles. They create macroeconomic effects in chains that affect several countries. Their influence is negative on companies. This causes insolvency, affects interest rate, growth, and other microeconomic factors.

The study indicates financial innovations as one of the determinants of crisis. This is related to the nature of innovation in the competitive financial services industry (^{Thakor, 2012}). More generally, ^{Mitchell (1941)} pointed out that the scope of business crises is due to the fact they are an intrinsic part of the corporate market.

Therefore, global financial crises make it difficult for companies to maintain their capital structures. It can make companies make the wrong decisions (^{Espinola, 2013}).

Operational definition of variables

Variables analyzed in the present study were divided between dependent and independent variables.

Dependent variables

Regarding the dependent variables, the research was based on previous studies such as ^{Shyam Sunder and Myers (1999)}, ^{Frank and Goyal (2003)}, ^{Leary and Roberts (2010)}, and ^{Espinola (2013)}. Debt indicators were treated at book value.

Total Debt: it indicates the percentage of short-term and long-term debt in relation to the company’s assets; the higher this percentage, the higher the company’s leverage level. When the TD index is high, it shows that there is great dependence on debt.

Short-Term Debt: it indicates the percentage of short-term debt in relation to the company’s assets.

Long-Term Debt: it indicates the percentage of long-term debt in relation to the company’s assets.

The three variables of indebtedness were chosen to evaluate whether there are specific drivers, according to the type of indebtedness. The financial literature has more research evaluating the long-term debt. Short-term indebtedness is little discussed in the literature.

Independent variables

The definition of independent variables was performed through an adaptation of empirical studies (Booth et al., 2001; ^{Brito; Corrar; Batistella., 2007}, ^{Nakamura et al., 2007}; ^{Bastos; Nakamura, 2009}; ^{Frank; Goyal, 2003}; ^{Espinola, 2013}).

Operating margin: it indicates the percentage between the operating incomes and revenues. According to the Pecking Order Theory, presented by ^{Brealey and Myers (2006)}, companies that show profits tend to use equity instead of debt. Therefore, their relation is inversely related to the level of indebtedness.

Size: expressed by the log of the revenues. By theory, larger firms tend to have low default risk, lowering the cost of borrowing. It has a direct effect on indebtedness.

Tangibility: it indicates the percentage of fixed assets in relation to total assets. Companies with more physical structures have greater tangibility.

ROE: it reflects the percentage of return on equity. Theories show that there is a discordance of analysis. According to ^{Jensen and Meckling (1976)}, because of Agency problems, managers seek to retain the company’s excess profits to make investments. Researchers show that companies committed to debt tend to have agency problems. Thus, according to the trade-off model, the most profitable companies should be the most leveraged. However, according to the Pecking Order Theory, described by ^{Myers (1984)}, excess of profit generates high indexes of reinvestment, which eliminates the necessity of searching for third-party capital. Therefore, the ratio is inverse to the level of leverage.

Growth: it indicates the percentage of growth of revenue over time. According to the Pecking Order Theory: the higher the revenue growth, the lower the need for debt.

Tobin’s Q ratio: this ratio was developed by ^{Tobin (1969)} and shows the relation between the market value, given by the expectation of future earnings, and the company’s book value. According to theories, companies with high indebtedness tend to add more market value.

Altman Z-score: The Z-score was developed by ^{Altman (1968)}. It is an indicator of credit risk. It is used to forecast bankruptcies of companies. It is calculated from multiple income and corporate balance values to measure the financial health of a company.

Hypotheses for normality periods

Data processing and statistical tests

In the study it was used a regression analysis with panel data, unbalanced. To this end, it was tested the hypothesis of two regression models, Fixed and Random, through the Hausman specification test, proposed by Hausman (1978). It was used Eviews 7 software for test and regression statistics. It has been widely used by many researchers in the area (^{Nakamura et al., 2007}, ^{Brito, Corrar; Batistella, 2007}; ^{Bastos; Nakamura, 2009}; ^{Frank; Goyal, 2003}; ^{Almeida et al., 2004}; ^{Espinola, 2013}).

Regression model:

where:

Validation of the regression model

The validation of the regression model was made by testing for the absence of multicollinearity and independence of residuals. Multicollinearity is the phenomenon of correlation between variables. Their evidence occurs when the R² of the regression is high and the variables are not significant. To this end, we analyzed the correlation matrix between the variables and the significance of the parameters. Residual dependence occurs when the regression’s errors are autocorrelated. To this end, the serial autocorrelation test proposed by ^{Durbin and Watson (1971)} was used.

Descriptive statistics

Table 6 shows the mean and standard deviation of the debt variables (dependent variables), which were calculated at book value, and Table 7 shows independent variables: operating margin, size, tangibility, ROE, growth, Tobin’s Q, and Altman Z.

Debt indicators show changes during the analyzed periods. It can be observed that in times of crisis the means are higher. It shows that debt was higher in this period. The standard deviation does not influence the analysis, since the sample was diverse and composed of companies from various industries and sizes.

In the descriptive statistics of the independent variables, it may be observed that all have changed in the compared periods. The operating margin shows a substantial change. It shows negative percentage in times of global crisis and in Brazilian economic recession. Considering the other indicators, several show lower mean in the crisis period; thus it may be considered that these indicators were affected by government and situational macro environment.

Estimated coefficients

Table 8 shows the results of the Hausman specification test statistics.

The results of Long-Term Debt in normality period and Total Debt in crisis period reject the null hypothesis. It shows that there are correlations between the regressors and error term. It indicates to estimate of models in the fixed form. The other models, however, accept the null hypothesis, indicating for an estimate of random models.

Tables 9, 10, and 11 consider the estimated coefficients of the proposed regression model of dependent variables (short-term debt, long-term debt, and total debt). In the Short-Term Debt model, the variables that have significance effects are those with p-value lower than 10%. Analyzing the normality period, operating margin, size, ROE, growth, and Altman Z were the variables that influenced the dependent variable. In the long-term debt it was noticed that operating margin, tangibility, ROE, growth, and Tobin’s Q were not significant. Finally, analyzing the total debt, operating margin, growth, and Tobin’s Q did not influence the regression model.

The behavior of indicators in crisis period was similar to the normality period. Both indicators were statistically significant. Independent variables with p-values greater than 10% are not statistically significant. In both periods the variables that most influenced the independent variables and had more differentiation between the periods were operating margin, growth, and Tobin’s Q.

When the variable becomes statistically insignificant in a period, this variable is discarded, as it is not possible to consider it in comparison.

In the long-term debt (for the normality period), size is determinant of capital structure; However, in crisis period, Tobins’s Q also becomes a determinant of long-term debt, confirming the findings of ^{Brito, Corrar, and Batistella (2007)}.

Hypotheses analysis

Through the analysis of the regression coefficients, it was possible to determine the relationship between the dependent variables and each of the explanatory factors (independent variables). Negative sign (-) shows a negative relationship, positive sign (+) shows a positive relationship and SI (statistically insignificant) means that there was no significance. Table 12 shows three columns of data: the first shows the expected relationship according to the hypotheses mentioned in the Methodology, the second shows the observed relationship in the normality period, and, finally, the third shows the observed relationship in the crisis period.

The result shows that the determinants of the capital structures are different between the normality period and the crisis period. Some variables did not reject the hypotheses mentioned in the Methodology. Following the premises of the Pecking Order Theory, the Operating margin variable, which measures the operating profit margin, should be negative.^{Brealey and Myers (2006)} point out that, according to the Pecking Order Theory, companies with higher operating margin tend to finance themselves (reinvestment) rather than through third-party capital.Table 12 shows that in Brazil this assumption is true, since the coefficient is negative. The greater the company’s operating margin, the lower the debt. However, in the normality period, the regression model showed a negative relationship: thus the hypothesis was not rejected. In fact, the model showed that there is a proportional relationship between the operating margin and debt indicators, but it was observed only in normality periods for a short-term debt. In other models statistics were not significant.

The Size variable also adhered well to the assumptions of theories. The great size of a company makes it easier to obtain financing, making the company more leveraged. These results are also in agreement with studies of ^{Ozkan (2001)} and ^{Gaud et al. (2005)}. However, in the models with Short-Term Debt variable this hypothesis is rejected. The same occurred with the Tangibility variable. Less tangible companies tend to have higher level of leverage. However, the tangibility parameter can assume positive or negative values, as presented by ^{Titman & Wessels (1988)}.

In relation to Tobin’s Q, which demonstrates the expected investment opportunity, it is observed that the positive values of Tobin’s Q corroborate the Pecking Order Theory, which states that the investment opportunity ratio should be positive in relation to the leverage level.

Although the Growth variable was not significant in crisis periods, the positive value in the normality period does not corroborate the Pecking Order Theory. The higher the revenue growth, the lower will be the dependence on debts. Therefore, this leads to a reduction in the leverage level.

The ROE variable also confirms the Pecking Order Theory, which states that more profitable companies are less indebted. This theory argues that companies should not seek external financing, without first exhausting their own resources. Therefore, higher operating margin tends to decrease leverage.

Similar to the ROE variable, the Altman Z variable is also in agreement with the theory. High scores reduce the level of credit risk and leverage. Therefore, if companies have more debts, they will have worse scores of credit risk.

It may be observed that the relationship of some dependent variables (short-term debt, long-term debt, and total debt) with the independent variables in the analyzed periods was different. Thus, the negative macroeconomic influences affect the determinants of the capital structures. ^{Espinola (2013)} used a sample of US companies and confirmed that in times of financial crisis determinants also changed, and operating margin, size, and tangibility had more significance than the others.

The initial hypotheses of variables (operating margin and ROE) were not rejected. Regarding the operating margin variable, ^{Myers (1984)}, ^{Donaldson (1961)}, and ^{Brealey and Myers (2006)} suggest that firms prefer raising capital, firstly from retained earnings, secondly from debt, and thirdly from issuing new equity. This relationship occurs because of the transaction costs involved in issuing equity, according to ^{Myers and Majluf (1984)}. However, in periods of economic expansion in Brazil, companies tend to use both equity and short-term debts. During this period, the Brazilian Development Bank (BNDES) provides credit at low costs.