nueva página del texto (beta)
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
Introduction
The capital structure choice is one of the most important strategic decisions made by managers, because of its wide implications for corporate performance and firm value. The influence of financial leverage on the economic value of the firm has been extensively studied, since the path-breaking work of Modigliani & Miller (1958) on the Irrelevance of Capital Structure. Ever since, a large number of empirical studies have tested MM’s Irrelevance of Capital Structure Theory in different social, economic and geographical settings, and in different time frames.
As a follow-up of their original contribution, Modigliani & Miller (1963) discuss a slightly different model in which they recognize the way corporate tax-shields may increase the value of a firm, as it increases its use of debt. The new model relaxes the original assumptions of a world with “no taxes”, perfect information and free from other frictions, and recognizes that the fiscal benefit of using debt is significantly larger than what they had originally anticipated. The main implication is that arbitrage opportunities will make the market value of firms within the same class to be a function not only of their expected returns, but also of the prevailing corporate tax rate and the firm’s debt level. MM clarify that their conclusion does not imply that firms should seek to use unlimited debt, and recommend managers to carefully ponder all other real-world costs associated to financial strategy. Later work proved that one of the most important of those real-world costs of using excessive debt is the risk of bankruptcy, reconciling the empirically observed use of debt, with the predictions of Modigliani & Miller’s analysis (see, for example, Senbet, & Seward, 1995, for a detailed survey on the subject).
During the early 1970’s there was relative consensus that an “optimal capital structure” must achieve a balance between the tax shield of debt, and the cost of financial distress. However, towards the end of that decade, Miller (1977) again raised the stakes by demonstrating that the fiscal advantage of corporate tax-shields could be completely offset by the effects of personal taxes.
A large number of studies have focused on the identification of the determinants of a firm’s capital structure. Some of the possible influential variables include, for example, the tangibility of the firm’s assets, the relative stability of cash flows, the intensity of competition, among others. Many of those studies have found confirmatory empirical evidence.1
In contrast, research on the influence of capital structure on a firm’s value shows only mixed results. Some works support Modigliani & Miller’s (1958) “irrelevance” of capital structure argument, while others find a positive relationship between leverage and firm value, as suggested by Modigliani & Miller (1963), and, still others report a negative relation between capital structure and firm value (e.g., Zingales & Rajan 1995). Although market imperfections (such as taxes, agency costs and asymmetric information) have been frequently cited to explain why capital structure is an important determinant of firm value, the nature of the relationship has not yet been completely clarified, and remains a subject of active discussions (Myers 1993, 2001).
It is important to mention that most of the published studies on the subject have been produced using data from the more advanced countries, where financial markets are deeper and more liquid, making emerging countries a special case that deserves more attention. The intention of this study is to contribute to a better understanding of the relationship between leverage changes and firms’ valuation in Mexico, an emerging economy.
By studying a sample of publicly traded Mexican non-financial firms, we find a positive and significant relationship between changes in different measures of leverage and changes in Tobin’s Q, used as a proxy for firm value. This is an already extensively documented stylized phenomenon; however, the original contribution that we introduce is on the way different characteristics of the firm affect the response of firms’ value to leverage, as discussed in more detail in the following sections.
The next section presents a brief literature review on the influence of debt utilization over firm value; the third section introduces the sample data, the econometric approach, and the results interpretation. The fourth section concludes, summarizes the findings, and suggests future research avenues.
Literature review
Modigliani & Miller (1958, 1963) established the theoretical framework that for decades has explained how does the use of debt impact a firm’s value. Since that seminal work, many other authors have contributed to develop the theory and to the empirical testing of its conceptual proposals. For example, the works of Stiglitz (1969 and 1988) offer evidence that capital structure is relevant for the determination of firm value, argues that the assumptions of Modigliani & Miller (1958) are not realistic, and suggest that capital structure affects firm value because of the different tax-rates paid on debt and equity income. Or, Leland & Pyle (1977) and Ross (1977), who suggest that firms use the announcements of capital structure changes to signal the quality of their future projects and, when market participants interpret those signals, they buy/sell stock, affecting the value of the firm.
De Angelo & Masulis (1980) implicitly recognize the influence of leverage on the value of the firm when they propose that the optimal capital structure is determined at the point where the marginal tax advantages of debt is equal to the marginal costs of financial distress, thus maximizing the market value of the firm.
Masulis (1983) was among the first to estimate the impact of a change in debt levels on firm value, and studies two forms of capital structure change: issuer exchange offers, and recapitalizations. The evidence reported confirms the existence of a positive relationship among firms’ value and leverage changes, which is consistent with optimal capital structure models, as well as with the idea that changes in debt levels convey information on firms’ value.
In one of the most influential works in the field Myers & Majluf (1984) show that if new investors are less informed than current investors about the quality of a firm, issuing new equity may cause an erosion in firm’s value because the market will infer the stock price is overvalued. These authors argue that new equity issues are interpreted by the market as “bad news”, so managers always prefer to fund new projects with debt, not with equity, implicitly imposing a positive relation between debt level and firm value.
Again, Fischer et al. (1989) refute Modigliani & Miller’s irrelevance proposition suggesting that an optimal dynamic capital structure can be found by considering the tax benefits of debt, bankruptcy costs, asset variability, the riskless interest rate, and the costs of recapitalization. In contrast, Zingales & Rajan (1995) and Frank & Goyal (2002) report a negative relation between the firm’s financial leverage and the market to book value ratio.
What can be learned from this rapid overview of the literature is that while some seminal authors associate greater financial leverage to higher firm value, others find an opposite relation between those variables, and still others find no relation at all. But, only in the “perfect” world of Modigliani and Miller (1958) is the value of the firm totally neutral to capital structure changes, so the latter case may be attributed to measurement bias or model definition issues.
In the context of emerging economies, there are again three different categories of empirical results. A number of studies find a negative relationship between debt utilization and firm value (e.g. Phung & Le 2013; Ivashkovskaya & Stepanova 2011; Ruan, Tian & Shiguang 2011; and, Tian & Zeitun 2007). Others find a positive relationship among emerging markets firms (e,g Mansourlakoraj 2015; Priya et al. 2015; Javeed & Azeem 2014; Olokoyo 2013; Chowdhury & Chowdhury 2010; and, Sarma & Rao 1969). And, still some others agree with the original Modigliani & Miller’s (1958), reporting that leverage does not have a statistically significant relationship with firm value (e.g. Hasan, Ahsan, Rajhans & Kaur 2014; Naceur, Ben & Goaied 2002; and, Krishnan & Moyer 1997).
Finally, some studies report mixed results. That seeming contradiction of terms may be explained either because of different measures of the value of the firm and/or its capital structure, or because there is a threshold on leverage’s influence on firm value. Some interesting examples include Jawad et al. (2015), Cheng & Tzeng (2014), Sheikh & Wang (2013), Lin & Chang (2011) and Iturriaga & Crisostomo (2010). A more comprehensive overview of these works is included in Table A1 in the Appendix.
Data, methodology and econometric results
By the end of 2014, 149 firms were listed in the Mexican Stock Exchange Market. Financial sector firms are excluded from our sample due to the peculiarities of their capital structure,2 leaving only 88 non-financial public firms. The sample is further reduced because only those firms whose financial statements and stock price are complete for the 2004 to 2014 period are included, leaving a final sample of 69 firms. The financial statements of those firms are retrieved from Capital IQ, and their stock prices, from Bloomberg. Almost 90% of the firms in our sample are part of the sectorial indices of the Mexican Stock Exchange. A complete list of the firms and some of their characteristics is included in Table A2 in the Appendix.
The variable of interest is the value of the firm, as proxied with Tobin’s Q. It is modeled using a set of explanatory variables that include financial leverage (extensively documented in the relevant literature as one of the main influences of firms’ valuation) and three additional control variables: size, profitability and risk. Understanding the relationship of changes in the independent variables (leverage, size, profitability and risk) with changes on the dependent variable (value of the firm) reveals interesting stylized facts of interest to different economic agents.
The dependent variable, Value of the firm for time is measured with Tobin’s Q3, as follows:
Financial leverage is measured using three different ratios: (i) the debt ratio4
(ii) following Tian & Zeitun (2007) and others, Debt to invested capital:
(iii) according to Modigliani & Miller (1958), Stiglitz (1988) and many others, the third measure of leverage is Debt to equity:
As mentioned, the results reported by numerous studies on the relationship between leverage and firm value are not conclusive.
Working Hypotheses
We postulate that leverage may generate increases in firm value due to debt’s relatively lower cost, but beyond a certain point, the cost of leverage increases as a result of potential bankruptcy costs, resulting in a negative impact on firm value. Our first working hypothesis is expressed as follows:
Hypothesis No. 1: Changes in Leverage have a positive relationship with changes in firm value for low leverage firms, and a negative relationship for high leverage firms.
In this study, the size of a firm is measured as the Log of Total Assets. Mixed results have been found with respect to the relationship between this variable and a firm’s value. According to some authors, there is a positive relationship as a result of the economies of scale enjoyed by large firms (e.g. Sheikh & Wang 2013; Tian & Zeitun 2007). Larger firms have more reputation, lower bankruptcy risk, lower information costs and less problems of information asymmetry; consequently, they also have higher value (Chen & Chen 2011).
Alternatively, a negative relationship may result due to the inefficiencies from the size of large firms, or diseconomies of scale (Jawad et al. 2015; Kouki & Said, 2011; Wippern, 1966). In the same way, Chaudhry & Sam (2014) found a negative relationship between firm size and stock returns. To test if that is the case in our sample firms, we postulate the following hypothesis:
Hypothesis No. 2: Changes in Size have a negative relationship with changes in firm value.
Profitability is measured in two ways: i) As in Friend & Lang (1988), Booth (2001) and Zingales & Rajan (1995), among others, the first measure of profitability is Return on Assets (ROA), calculated as EBIT/Assets. ii) According to Krishnan & Moyer (1997), the second measure of profitability is the return on invested capital, ROIC, defined as EBIT minus taxes divided by the sum of debt plus equity. A positive relationship between this variable and Firm Value is expected a priori (Kouki & Said, 2011). Likewise, Chen & Chen (2011) found the greater the profitability of a firm, the more the assignable profit, and the higher the value of the firm. Kurniasari & Warastuti (2015), Yang et al. (2010), Haugen & Baker (1996), Varaiya, Kerin & Weeks (1987) also found a positive relationship between profitability and firm value. In this study we test:
Hypothesis No. 3: Changes in Profitability have a positive relationship with changes in firm value.
According to Krishnan & Moyer (1997), risk is conceptualized as the variability of profits. In this study, it is operationalized as the Log of the standard deviation of EBITDA during the previous five years.
Different theories explain the relation between firm value and risk. For example, Modigliani & Miller (1958) postulate that, in perfect markets, the firm’s value is independent from risk. Empirical evidence has shown that the perfect market hypothesis does not hold, and that risk can lead to deadweight costs (Nocco & Stulz 2006). The agency theory postulates there is a positive relationship that results from risk-averse behavior (Ross 1973). This theory is supported by several empirical works (e.g., Fletcher 2000; Maurer 2008) that find a positive relationship between beta (CAPM) as a measure of risk, and stock returns which, eventually, make stocks more valuable. Similarly, Abdel-azim (2015) found a positive relationship between risk, measured by beta (CAPM), and Tobin’s Q, used as a proxy for firm value. By contrast, there are several empirical works that have found a negative relationship between risk and return. For example, Bowman’s paradox states that decision makers are risk-seeking and assume higher risk for lower return (Bowman, 1982). Bowman’s paradox was confirmed by other empirical works (e.g., Fama and French, 1992; Andersen et al., 2007). We test for risk’s influence on firm value with Hypothesis No. 4:
Hypothesis No. 4: Changes in Risk have a negative effect on changes in firm value.
Methodological Design
Following Alonso et al. (2005) the median value of the distribution of each explanatory variable is utilized to subdivide the sample in two groups:
Low and high leverage: With median values of 47.4% for the debt ratio, 30.4% for debt to invested capital and 45.2% for the debt to equity ratio.
Small and big size: With a median of Log of Assets of 3.04.
Low and high profitability: ROA with median value of 4.7% and ROIC, with median of 6.51%.
Low and high risk: The median value for the log of the standard deviation of EBITDA is 1.44.
All firms with lower values than the corresponding median, conformed the samples of “low” levels, the rest of the firms were included in the samples of “high” levels. Low levels in all subsamples consisted of 34 firms and the high levels included 35 firms.
Given the characteristics of the sample (short period of observations and large number of entities), the recommended econometric approach is Panel Data Analysis, including Pooled Ordinary Least Squares (OLS), Fixed Effects (FE) and Random Effects (RE). Our interest lies in the measurement of changes in firms’ value (as measured by first-differences in Tobin’s Q), with respect to the explanatory variables (also in first-differences). The basic model is expressed as follows:
Where:
dVit: Change in Value of the i firm for time t
α = Intercept or the firm-varying linear coefficient
dLEVER: Change in Leverage
dSIZE: Change in Size
dPROFIT: Change in Profitability
dRISK: Change in Risk
β1 , β2 , β3, β4 =Angular coefficients to be estimated
ε it = Disturbance term or residual
With i = 1, … , N, the number of firms; t = 1, … , T, the time periods, and εit, the error term, expected to be normally distributed with zero mean and constant variance:
The model yields unbiased and efficient estimates of the parameters, since it minimizes the loss of degrees of freedom. The panel data approach allows for the control of the “unobservable constant heterogeneity” (Arellano & Bover 1990) to control for variables that cannot be measured or directly observed (like organizational culture factors or differences in business practices across firms) and for variables that change over time but not across firms (i.e. national policies, federal regulations, international agreements, etc.). Furthermore, the dynamic dimension of panel data supports testing long time adjustment processes that determine the sensitivity of firm value to the explanatory variables changes (Alonso et al., 2005).
From the basic model we test three specific variations for each of the nine samples (considering the different measures of leverage and profitability):
Model 1:
Where:
dLEVER1: Change in Leverage defined as D/A
dPROFIT1: Change in Profitability defined as ROA
Model 2:
Where:
dLEVER2: Change in Leverage defined as D/(D+E)
dPROFIT2: Change in Profitability defined as ROIC
Model 3:
Where:
dLEVER3: Change in Leverage defined as D/E
dPROFIT2: Change in Profitability defined as ROIC
Table A3 in the Appendix contains the descriptive statistics of all the variables. It is interesting to notice that dV it has a mean of around -0.12 in most sub-samples meaning that, on average, Mexican non-financial public firms value decreased 12 percentage points annually during 2004-2014.
The three measures of leverage indicate that the sample of Mexican Public firms did not have important changes in capital structure on average, during the period of observation, i.e., the average change in the three measures was very close to zero.
The Change in the Size of Firms suggests that the average firm in the sample increased in size by around 2% annually.
The two measures of Profitability reveal that, during the period of 2004-2014, sample firms reduced their profitability between 1% to almost 20% in the different sub-samples. The smallest reduction was for the low leverage sample and the largest was for the high leverage sample. Finally, Risk increased by an average of 1% annually during 2004-2014.
The Panel Analysis model is estimated using Pooled Ordinary Least Squares (OLS), Fixed effects (FE) and Random effects (RE) regressions.
The Pooled model is the most restrictive, since it specifies constant coefficients. If the model is correctly specified and the independent variables are not correlated with the regression error, then it can be consistently estimated using Ordinary Least Squares.
The Fixed Effects Model (FE) considers the unique characteristics of every unit (firm) from cross section data, and captures them with different intercepts for each entity; however, the angular coefficients are constant or fixed across entities. In the Fixed Effects Model, the unobserved variables do not change over time, so it can be assumed that any change in the dependent variable is due to influences other than the fixed characteristics.
The Random Effects Model (RE) assumes that the entity’s error term is uncorrelated with the explanatory variables, which allows for time-invariant variables to play a role as independent variables. Using RE it is possible to generalize the inferences beyond the sample used in the model, something that cannot be done with FE.
To choose among the three models: Pooled Ordinary Least Squares, Fixed Effects or Random Effects, two tests are used: (i) the Redundant Fixed Effects Test, to decide between Pooled Ordinary Least Squares and Fixed Effects; and, (ii) the Hausman Test to decide which of the two, Fixed Effects or Random Effects, is more appropriate.
Results and Analysis
We first present the results of the panel regressions for the complete sample under the three models in Table 1, below. Only the best adjustment model is reported in the table; however, a complete report can be obtained upon request.
Table 1 Regression Results: Whole Sample.
| Dependent Variable: | dV | ||
|---|---|---|---|
| Model | 1 | 2 | 3 |
| Leverage | D/A | D/(D+E) | D/E |
| Profitability | ROA | ROIC | ROIC |
| Preferred Estimation Method | FE | FE | FE |
| dLEVER | 0.73919 | 0.2860 | 0.004488 |
| (5.815054)*** | (2.677)*** | (0.598405) | |
| dSIZE | -2.991714 | -3.344 | -3.297504 |
| (-22.35449)*** | (-20.563)*** | (-20.28778)*** | |
| dPROFIT | 0.008323 | 0.0013 | 0.001382 |
| (2.321158)** | (1.8703)* | (2.060812)** | |
| dRISK | -0.171891 | -0.1322 | -0.120659 |
| (-2.714313)*** | (-1.81432)* | (-1.65043)* | |
| CONS | -0.026838 | -0.0355 | -0.035736 |
| (-2.63327)*** | (0.0041)*** | (-2.88835)*** | |
Source: Own elaboration with data retrieved from Capital IQ and panel regressions in Eviews.
Note: dV: Change in firms’ Value, measured with Tobin’s Q. dLEVER: Change in Leverage (three measures). dSIZE: Change in the Size of the firm. dPROFIT: Change in Profitability (two measures). dRISK: Change in Risk. *** = Significant at a 1% level; ** = Significant at a 5% level and * = Significant at a 10% level. Numbers in parentheses are asymptotic t‐values.
The relationship between changes in leverage and changes in firm value in the three models is positive and highly statistically significant in models 1 and 2, but loses significance at conventional levels for model 3. According to model 1, if financial leverage increases by 1%, the Firm Value will increase in 0.739%, consistent with the results reported in several works (e.g. Mansourlakoraj 2015; Priya 2015; Javeed & Azeem 2014; Olokoyo, 2013; Chowdhury & Chowdhury 2010; Masulis 1983; Ruland & Zhou 2005; Sarma & Rao 1969; and Wippern 1966). Changes in size also have a negative and statistically significant coefficient for the three models, confirming the theory that, as size increases, the firm experiences inefficiencies (diseconomies of scale) that will cause its value to diminish. The results also indicate that there is a very small positive, but significant relationship between changes in profitability and changes in the value of the firm. Likewise, there is a significant negative relationship between risk and firm value in all three models.
The results for the two sub-samples that are classified according to the level of financial leverage, i.e., low and high leverage firms, are presented in Table 2.
Table 2 Subsample Regression Results: Low Leverage vs High Leverage Firms.
| Dependent Variable: dVit | ||||||
|---|---|---|---|---|---|---|
| Low Leverage Firms | High Leverage Firms | |||||
| Model | 1 | 2 | 3 | 1 | 2 | 3 |
| Leverage | D/A | D/(D+E) | D/E | D/A | D/(D+E) | D/E |
| Profitability | ROA | ROIC | ROIC | ROA | ROIC | ROIC |
| Pref. Estim. Method | FE | FE | FE | RE | FE | FE |
| dLEVER | 0.8571 | 0.4771 | 0.2275 | 0.7162 | 0.2294 | 0.0033 |
| (5.9723)*** | (4.0194)*** | (3.6604)*** | (3.6587)*** | (1.4073) | (0.3442) | |
| dSIZE | -3.3613 | -3.4425 | -3.4218 | -2.8463 | -3.3077 | -3.2845 |
| (-23.8193)*** | (-22.7593)*** | (-22.9346)*** | (-12.9793)*** | (-12.2728)*** | (-12.1046)*** | |
| dPROFIT | 0.0052 | 0.0054 | 0.0050 | 0.0147 | 0.0012 | 0.0013 |
| (1.7187)* | (2.2977)** | (2.2191)** | (1.8557)* | (1.3497) | (1.5067) | |
| dRISK | -0.1676 | -0.1601 | -0.1847 | -0.1739 | -0.1294 | -0.0873 |
| (-2.5502)** | (-2.4594)** | (-2.8566)*** | (-1.6557)* | (-1.0249) | (-0.6830) | |
| CONS | -0.0165 | -0.0261 | -0.0252 | -0.0305 | -0.0434 | -0.0432 |
| (-1.6463) | (-2.5636)** | (-2.4614)** | (-0.9428) | (0.0517)* | (-1.9420)* | |
Source: Own elaboration with data retrieved from Capital IQ and panel regressions in Eviews.
Note: dV: Change in firms’ Value, measured with Tobin’s Q. dLEVER: Change in Leverage (three measures). dSIZE: Change in the Size of the firm. dPROFIT: Change in Profitability (two measures). dRISK: Change in Risk. *** = Significant at a 1% level; ** = Significant at a 5% level and * = Significant at a 10% level. Numbers in parentheses are asymptotic t-values.
For the three models the relationship between change in leverage and change in firm value remains positive and highly statistically significant among low financial leverage firms; but for high leverage firms that relationship is only significant for model 1. In the case of the control variables, the relationship with firm value are all statistically significant for low leverage firms, but not so for high leverage firms; only the first model’s control variables are statistically significant in all cases.
In the case of models 2 and 3 for high leverage firms, not only does leverage lose statistical significance but, while the coefficients of all the control variables keep the same sign changes in profitability and in risk lose their statistical significance. The change in size shows a negative and statistically high significant relationship with firm value in all the models.
Table 3 reports the estimation results for the sub samples obtained using the size of the firm as the analytical variable:
Table 3 Subsample Regression Results: Small Size vs. Large Size Firms
| Dependent Variable: dV | ||||||
|---|---|---|---|---|---|---|
| Small Size Firms | Large Size Firms | |||||
| Model | 1 | 2 | 3 | 1 | 2 | 3 |
| Leverage | D/A | D/(D+E) | D/E | D/A | D/(D+E) | D/E |
| Profitability | ROA | ROIC | ROIC | ROA | ROIC | ROIC |
| Pref. Estim. Method | FE | FE | FE | RE | FE | FE |
| dLEVER | 0.5970 | 0.3837 | 0.0144 | 1.0139 | 0.1657 | 0.0024 |
| (3.1609)*** | (2.7758)*** | -0.6021 | (6.2806)*** | (1.0153) | (0.2978) | |
| dSIZE | -2.7233 | -2.7747 | -2.6903 | -3.4289 | -4.2985 | -4.2782 |
| (-13.5376)*** | (-13.3109)*** | (-12.9076)*** | (-20.3983)*** | (-17.6897)*** | (-17.6693)*** | |
| dPROFIT | 0.0070 | 0.0044 | 0.0030 | 0.0062 | 0.0010 | 0.0011 |
| (1.4366) | (1.2720) | (0.8683) | (1.1173) | (1.4292) | (1.5875) | |
| dRISK | -0.2517 | -0.2640 | -0.2295 | -0.0546 | 0.0388 | 0.0393 |
| (-2.4308)** | (-2.5753)** | (-2.2291)** | (-0.7354) | (0.7013) | (0.3883) | |
| CONS | -0.0361 | -0.0462 | -0.0476 | -0.0102 | -0.0073 | -0.0071 |
| (-2.2240)** | (0.0052)*** | (-2.8707)*** | (-0.5630) | (-0.3051) | (-0.3012) | |
Source: Own elaboration with data retrieved from Capital IQ and panel regressions in Eviews.
Note: dV: Change in firms’ Value, measured with Tobin’s Q. dLEVER: Change in Leverage (three measures). dSIZE: Change in the Size of the firm. dPROFIT: Change in Profitability (two measures). dRISK: Change in Risk. *** = Significant at a 1% level; ** = Significant at a 5% level and * = Significant at a 10% level. Numbers in parentheses are asymptotic t‐values.
The relationship between leverage and firm value continues to be positive for small and large firms under the three models, but among the latter it is significant in model 1 only. Change in size and change in firm value have a negative and high statistically significant relationship in the three models, although it is weaker for small firms and stronger for large firms, relative to the complete sample. For both, small and large firms, and for the three models, the relationship between the change in profitability and firm value continues to be positive, but not significant. Regarding the relationship between risk and firm value, it is negative for both, small and large firms, under the three models, but completely loses statistical significance for large firms. For smaller firms, that relationship is stronger relative to the complete sample, and remains statistically significant for the three models. This means that the negative effect of size on firm value is more intense in smaller firms. Table 4 reports the Low and High Risk sub samples’ estimates.
Table 4 Subsample Regression Results: Low Risk vs. High Risk Firms
| Dependent Variable: dV | ||||||
|---|---|---|---|---|---|---|
| Low Risk Firms | High Risk Firms | |||||
| Model | 1 | 2 | 3 | 1 | 2 | 3 |
| Leverage | D/A | D/(D+E) | D/E | D/A | D/(D+E) | D/E |
| Profitability | ROA | ROIC | ROIC | ROA | ROIC | ROIC |
| Pref. Estim. Method | FE | FE | FE | RE | FE | FE |
| dLEVER | 0.5685 | 0.3595 | 0.0115 | 1.0498 | 0.2076 | 0.0029 |
| (2.9089)*** | (2.6290)*** | (0.4882) | (6.5431)*** | (1.2444) | (0.3594) | |
| dSIZE | -2.8032 | -2.8258 | -2.7522 | -3.2908 | -4.1537 | -4.1208 |
| (-13.7703)*** | (-13.6589)*** | (-13.2981)*** | (-20.0433)*** | (-17.0356)*** | (-16.9704)*** | |
| dPROFIT | 0.0066 | 0.0027 | 0.0007 | 0.0068 | 0.0011 | 0.0012 |
| (0.8959) | (0.5603) | (0.1499) | (1.8711)* | (1.5697) | (1.7529)* | |
| dRISK | -0.2584 | -0.2588 | -0.2263 | -0.0720 | 0.0147 | 0.0132 |
| (-2.4719)** | (-2.5670)** | (-2.2375)** | (-0.9866) | (0.1424) | (0.1274) | |
| CONS | -0.0337 | -0.0430 | -0.0445 | -0.0156 | -0.0157 | -0.0156 |
| (-2.0616)** | (-2.669)*** | (-2.7389)*** | (-0.8570) | (-0.6498) | (-0.6470) | |
Source: Own elaboration with data retrieved from Capital IQ and panel regressions in Eviews.
Note: dV: Change in firms’ Value, measured with Tobin’s Q. dLEVER: Change in Leverage (three measures). dSIZE: Change in the Size of the firm. dPROFIT: Change in Profitability (two measures). dRISK: Change in Risk. *** = Significant at a 1% level; ** = Significant at a 5% level and * = Significant at a 10% level. Numbers in parentheses are asymptotic t‐values.
The relationship between Leverage and Firm Value remains positive but is not statistically significant three of the six models (model 3 of the Low Risk sub sample, and models 2 and 3 of the High Risk sub sample). The negative and highly statistically significant relationship between size and firm value in all six models is less intense among Low Risk firms and stronger among High Risk firms, in comparison to the whole sample. The positive and small size of the coefficients for the relationship between profitability and firm value remains, but loses statistical significance in most cases. A negative sign is preserved in the relationship between risk and firm value, but loses statistical significance for the High Risk sub sample. Finally, Table 5 presents the Low and High Profitability sub samples’ results.
Table 5 Subsample Regression Results: Low Profitability vs. High Profitability Firms
| Dependent Variable: dV | ||||||
|---|---|---|---|---|---|---|
| Low Profitability | High Profability | |||||
| Model | 1 | 2 | 3 | 1 | 2 | 3 |
| Leverage | D/A | D/(D+E) | D/E | D/A | D/(D+E) | D/E |
| Profitability | ROA | ROIC | ROIC | ROA | ROIC | ROIC |
| Pref. Estim. Method | RE | RE | RE | RE | FE | FE |
| dLEVER | 0.7752 | 0.4367 | 0.0054 | 0.8963 | 0.2064 | 0.0140 |
| (8.0049)*** | (5.4371)*** | -1.1987 | (3.6372)*** | (0.9955) | (0.3147) | |
| dSIZE | -2.1776 | -2.3734 | -2.3277 | -4.2078 | -4.7005 | -4.6476 |
| (-21.6267)*** | (-19.9095)*** | (-18.8589)*** | (-16.8112)*** | (-14.9584)*** | (-14.9907)*** | |
| dPROFIT | 0.0101 | 0.0026 | 0.0001 | 0.0073 | 0.0010 | 0.0012 |
| (2.4999)** | (0.8950) | (0.0295) | (1.4095) | (1.2302) | (1.4112) | |
| dRISK | -0.0571 | -0.1324 | -0.0905 | -0.2429 | -0.0976 | -0.1026 |
| (-1.1520) | (-2.2883)* | (-1.5193) | (-2.1200)** | (-0.7564) | (-0.7945) | |
| CONS | -0.0033 | -0.0088 | -0.0094 | -0.0195 | -0.0329 | -0.0336 |
| (-0.2368) | (-0.6035) | (-0.6420) | (-0.6418) | (-1.4243) | (-1.4497) | |
Source: Own elaboration with data retrieved from Capital IQ and panel regressions in Eviews.
Note: dV: Change in firms’ Value , measured with Tobin’s Q. dLEVER: Change in Leverage (three measures). dSIZE: Change in the Size of the firm. dPROFIT: Change in Profitability (two measures). dRISK: Change in Risk. *** = Significant at a 1% level; ** = Significant at a 5% level and * = Significant at a 10% level. Numbers in parentheses are asymptotic t‐values.
In this case, the sign of the coefficient for change in Leverage is positive in all six models, but only significant in Low Profitability firms’ models 1 and 2, and in model 1 of the High Profitability sub sample. Changes in Size are highly statistically significant and have a negative sign in all six models, but the magnitude of the Low Profitability models is about half the size of the High Profitability models, i.e., as High Profitability firms grow in size, the market penalizes possible diseconomies of scale twice as much than in the case of Low Profitability firms. The relation between changes in Profit and changes in Risk, versus Firm Value is not significant in most of the regressions, but in all cases preserves the signs of the whole sample.
Our results are fully consistent with the Trade-Off Theory, and imply that managers can find an optimal capital structure (one that maximizes firm value), as long as the advantages of using debt (greater tax shield and lower agency costs) are more than the costs of debt (cost of issuing, bankruptcy costs, and agency costs of debt). Accordingly, firm value may be augmented by increasing leverage until the marginal benefits from additional debt are equal to the marginal costs of the firm’s bankruptcy costs.
All the models suggest the presence of a significant positive relationship between changes in leverage and changes in value. However, the only model that is highly statistically significant in all cases is model 1, which uses the debt ratio to measure leverage. Model 2, with the ratio of debt to invested capital as a measure of leverage, loses statistical significance for all low-debt subsamples. In the case of model 3, using the debt to equity ratio as a proxy for leverage, most of the explanatory variables lose their statistical significance except for Size, which was the variable with the highest and most consistent statistical significance in all of the models and samples. Since in model 1 (which uses D/A as a measure for Leverage) the Leverage coefficient has a positive sign and a high statistical significance for the whole sample, for all the sub samples, and, at the same time, shows statistical significance for the coefficients for most other explanatory variables, our conclusions in the following section refer to it.
Conclusions
This paper tests the effects of changes in leverage on the market value of Mexican publicly traded firms using three econometric estimation methods on the whole sample and eight different sub samples. Our results reject the “Irrelevance Theory” of Modigliani & Miller (1958) and indicate that capital structure changes certainly affect the market value of our sample firms; however, they are fully consistent with Modigliani & Miller (1963). They are also congruent with DeAngelo & Masulis (1980), who suggest there is an “optimal capital structure” since in our estimation’s positive relation between leverage and firm value proves to be stronger among firms with low levels of leverage, while the positive effect of leverage is weaker for firms that show high levels of financial leverage; i.e., there seems to be a threshold beyond which leverage no longer increases firm value, suggesting the existence of an optimal capital structure.
Our empirical evidence also supports Myers & Majluf (1984) who postulate that, due to information asymmetries, seasoned equity offers (SEOs) tend to be underpriced, as the market perceives new issues of equity as a proxy indicator of “bad news”, so the use of more equity as a proportion of total funds depresses the stock price. In an emerging market environment, the asymmetry of information between managers and new investors is even more serious than in more developed markets, so one should see an even more unfavorable valuation of firms that prefer equity to debt financing. That seems to be the case for our Mexican firms sample.
Segmenting by size, estimation results indicate the relationship between changes in Leverage and Firm Value is weaker among Small Size firms, and stronger among Large Size firms. Segmenting by Profitability yields similar results, the value (proxied by Tobin’s Q) of Low Profitability firms is less affected by Leverage than is the case for High Profitability firms.
The main findings of this work are summarized as follows:
In general, changes in leverage have a positive effect on firm value for both, low and high leverage firms, although the benefits are greater for low leveraged firms, confirming Hypothesis No. 1. We find no evidence that suggests that the use of leverage decreases the value of firms. These results are consistent with the findings of: Sarma & Rao (1969), Javeed & Azeem (2014), and Priya & Protheepan (2015). At the same time, they contradict the findings reported for emerging markets in: Tian & Zeitun (2007), Phung & Le (2013) and Ruan, et al. (2011), and may be attributed to the tax shield associated with leverage, as well as the lower cost of capital from debt.
For all nine samples our results indicate that size has a negative impact on relative firm value (as measured by Tobin’s Q), confirming Hypothesis 2. This negative relationship is stronger for big firms. These findings are similar to those reported in Jawad et al (2015), but contradict the findings of Sheik & Wang (2013) and Tian & Zeitun (2007). We interpret this finding as an expression of diseconomies of scale reflected on market valuation.
Changes in profitability have a positive effect on firm value. However, the profitability coefficient was the lowest among all the estimated coefficients in all nine sub samples. Therefore, we conclude that, in the case of Mexican Public firms, the effect of profitability on firm value is positive but not strong, i.e., Hypothesis 3 cannot be rejected. Loncan & Caldeira (2014) similarly report a positive effect of profitability on firm value among Brazilian firms. It could be argued that the accounting measure of profitability is not the preferred measure of value creation of a firm among Latin American investors.
Changes in risk have a negative influence on changes in the firm value of Mexican non-financial public firms, confirming our Hypothesis 4. Lin & Cheng (2011) had the same results in a study on Taiwanese firms.
Accordingly, our reported evidence suggests that leverage positively influences firm value, and the “decreasing improvements” in value as debt levels increase could suggest an optimal capital structure, providing evidence that supports the models of Cheng & Tzeng (2014), Kim (1978) and Stiglitz (1972).
These finding have relevant implications for policymakers, academicians and practitioners. For policymakers, it shows the importance of promoting the development of an efficient capital market, and of the transparency of corporate financial laws that make the cost of bankruptcy an important issue to consider when making capital structure decisions. For academicians, this study’s results shed light on the different effects of leverage for different types of firms, and also on the importance of capital structure decisions in emerging markets. For practitioners, our results may be of help to make better capital structure decisions considering the fundamental characteristics of their firms and, finally, for investors, the relationship between leverage and other characteristics of firms can help portfolio managers optimize their selection of investment assets.
