资本结构文献

ctu f ti l li le or rag firm s on c char rmina oss tim financing policies. These authors (including Booth et al., 2001; Ban- cel and Mittoo, 2004; Antoniou et al., 2008; Beck et al., 2008; de Jong et al., 2008) suggest that, along with firm characteristics, country-specific factors may also influence firm capital structure. These studies compare the capital structure of firms from different countries, taking into account factors such as gross domestic prod- uct (GDP), development of stock markets, levels of investor protec- tion, etc. Furthermore, some papers (e.g., Burgman, 1996; Chen analysis – with maximum likelihood estimation in order to assess all levels simultaneously. As we shall discuss further, the capital structure determinants can be nested in at least three levels: level 1 (time), level 2 (firm characteristics) and level 3 (the industry/ country interaction). In this context, we assume that the character- istics of higher levels may influence the characteristics of lower levels. For example, firms (lower level) working in a given industry (higher level) have similar patterns of behavior and, hence, would have similar leverage ratios. Thus, such firms will tend to have a strong within-cluster correlation. However, these firms may differ from other firms of different industries, leading to significant ⇑ Corresponding author. Tel.: +55 11 8346 5722; fax: +55 11 2206 1745. Journal of Banking & Finance 35 (2011) 358–371 Contents lists availab k w. E-mail addresses: kayo@usp.br (E.K. Kayo), hkimura@mackenzie.br (H. Kimura). Levy, 2003), albeit oftentimes converging on relatively stable cap- ital structures (Lemmon et al., 2008), which suggests the existence of an optimal level of leverage. Incidentally, the issue of a given capital structure that may increase the shareholder value is one of the most important discussions in the finance field, both theo- retically and empirically. Since Modigliani andMiller’s (1958) irrel- evance propositions, we have been witnessing the development of many theoretical points of view in this arena. Several studies analyze the role of countries and industries on Although the majority of capital structure studies include dummy variables representing different industries, only a few include vari- ables that characterize – but do not classify – each industry. Nota- ble exceptions are the studies of Simerly and Li (2000) and MacKay and Phillips (2005). Our paper seeks to analyze the influence of time-, firm-, indus- try- and country-level determinants of leverage of firms from 40 countries. Because of the multilevel nature of these determinants, we use hierarchical linear modeling (HLM) – also called multilevel Capital structure Hierarchical analysis Firm-level determinants Industry-level determinants Country-level determinants 1. Introduction The preponderance of the studie focuses on the analysis of certain firm ability, tangibility, size, etc. – as dete tion, capital structure may vary acr 0378-4266/$ - see front matter � 2010 Elsevier B.V. A doi:10.1016/j.jbankfin.2010.08.015 apital structure mainly acteristics – e.g., profit- nts of leverage. In addi- e (e.g., Korajczyk and et al., 1997; Mansi and Reeb, 2002; Desai et al., 2004) compare financing policies of multinational firms versus domestic firms based on the argument that global factors might influence financial leverage. If, on the one hand, it is easy to find studies that analyze firm/country characteristics as determinants of capital structure, on the other hand, the literature often neglects the role of industry. F30 G32 Keywords: financial behavior between firms of developed and emerging countries. � 2010 Elsevier B.V. All rights reserved. Hierarchical determinants of capital stru Eduardo K. Kayo a,⇑, Herbert Kimura b aUniversidade de São Paulo, São Paulo, Brazil bUniversidade Presbiteriana Mackenzie, São Paulo, Brazil a r t i c l e i n f o Article history: Received 3 February 2010 Accepted 16 August 2010 Available online 20 August 2010 JEL classification: a b s t r a c t We analyze the influence o First, we apply hierarchica We find that time and firm and random coefficients in characteristics on firm leve try and country-levels on Journal of Ban journal homepage: ww ll rights reserved. re me-, firm-, industry- and country-level determinants of capital structure. near modeling in order to assess the relative importance of those levels. vels explain 78% of firm leverage. Second, we include random intercepts der to analyze the direct and indirect influences of firm/industry/country e. We document several important indirect influences of variables at indus- determinants of leverage, as well as several structural differences in the le at ScienceDirect ing & Finance elsevier .com/locate / jbf (ii) panel regressions lead to problems of correlated residuals across years. In addition to addressing problems of time and firm anki effects, HLM also allows us to include the effects of industry and country in our analysis. In this context, our objective is twofold. First, we assess the rel- ative importance of each of these levels on the variance of firm leverage. We achieve this objective by using an empty model (i.e., without covariates) through which we find that the levels of time and firm are responsible for the majority of firm leverage var- iance. Second, we extend this basic model with the inclusion of random-intercepts and random-slopes in order to analyze, respec- tively, the direct and indirect influences of the characteristics of firm, industry and country on firm leverage. In pursuing our second objective, while we include traditional determinants of leverage at both firm and country levels, we also analyze three important characteristics of industries, i.e., munificence, dynamism and con- centration (Herfindahl–Hirshman index). While we can find some studies examining the relationship between industry concentra- tion and firm leverage (e.g., MacKay and Phillips, 2005), the study of munificence and dynamism is somewhat novel in the literature of capital structure. Munificence represents the abundance of re- sources in a given industry; dynamism is the instability or volatil- ity of that industry (Boyd, 1995). To the best of our knowledge, our study is the first to analyze the direct influence of these variables on firm leverage. Simerly and Li (2000), while they analyze indus- try dynamism, it is only as a moderator variable of leverage on firm return-on-assets. Moreover, in the present study we expand the discussion in order to includemacroeconomic determinants of firm leverage. Finally, we build on previous studies (e.g., de Jong et al., 2008) with the discussion of the indirect influences of country- and industry-specific variables on firm-specific determinants of leverage. The main results of our paper show that a significant part of the leverage variance – empty model reports nearly 42% – is due to intrinsic firm characteristics. Second, time-level is also responsible for a relevant part of leverage (36%). The industry-level character- istics, in turn, account for nearly 12% of leverage variance, and country-level, for only 3%. The remaining 7% of leverage variance is due to combined industry/country effects. Although the vari- ances attributable to industry and country are relatively low, this is not tantamount to stating that industry- and country-level fac- tors are unimportant. In fact, some of these factors are quite impor- tant to explain the firm leverage. The industry characteristics of munificence, dynamism and concentration, for instance, influence leverage significantly. We organize the remaining of the paper as follows. The next section sets out the theoretical discussion on determinants of cap- ital structure at the levels of firm, industry and country. Following, we describe the methodological procedures regarding data gather- ing, sampling, construction of measures and empirical models. The subsequent section shows our empirical results. The final section concludes our paper with the main theoretical and managerial implications of our results. 2. Determinants of capital structure 2.1. Firm-level determinants differences across clusters. HLM is able to mitigate the econometric problems raised by Fama and French (2002) regarding the charac- teristics of the data. They state that (i) the use of cross-section regressions ignores the correlations of residuals across firms and E.K. Kayo, H. Kimura / Journal of B Concerning firm-level determinants of leverage, three main the- oretical approaches are particularly important: the trade-off, the agency and the pecking order hypotheses. These theories, in contrast to Modigliani and Miller’s (1958) assumption of a perfect market, suggest that several factors may determine firm leverage, either firm-internal or firm-external. A particular factor might be positive or negative depending on the theoretical lens. Therefore, we analyze five firm-level determinants of capital structure: growth opportunities, profitability, distance from bankruptcy, size and tangibility. Competitive theories and predictions involve the relationship between growth opportunities and leverage. While the agency the- ory suggests a negative relationship between growth opportunities and leverage, the pecking order theory predicts this relationship is positive. The agency theory explanation for the negative relation- ship is based on the disciplinary role that debt can play in mitigat- ing the opportunistic behavior of managers. This kind of behavior is more pronounced when firm free cash flow is high. When the firm is in a high growth phase and investment opportunities with positive net present value are abundant, free cash flow is low and manager/shareholder conflicts are less intense. In this phase, debt may lead to underinvestment problems (Stulz, 1990) which ex- plains why firms tend to show high levels of equity rather than leverage. On the other hand, when growth opportunities are scarce, excess free cash flowmay give rise to typical agency problems such as adverse selection, moral hazard and excessive perquisites. In this scenario, debt plays an important role in motivating managers to be more efficient (Jensen, 1986). D’Mello and Miranda (2010) show recent empirical evidence that debt issues decrease excessive cash ratios, lowering abnormal capital expenditures and increasing the firm’s value. However, there are evidences that this disciplinary role of debt is more likely to occur in the absence of managerial entrenchment (Zwiebel, 1996; de Jong and Veld, 2001). A negative relationship between growth opportunities and leverage may also reflect the uniqueness – or specificity – of firms, especially their intangibility. Bah and Dumontier (2001) and O’Brien (2003), for instance, show that companies with higher re- search & development (R&D) and advertising expenses – both proxies of intangibility – have smaller levels of leverage. O’Brien (2003) argues that the low leverage in intangible-intensive compa- nies is due to equity flexibility, ensuring the accomplishment of investment-goals in R&D, the launch of new products and the acquisition of other companies in order to increase the knowledge base. The long period of maturation of such investments makes leverage an inappropriate source of funding. Accordingly, Brown et al. (2009) state that R&D-intensive firms (e.g., young public high-tech firms) mostly rely on internal or external equity to fund their projects, since they are subject to high asymmetric informa- tion, high uncertain returns, and low collateral value. Nevertheless, growth opportunities can also correlate positively with leverage, according to the pecking order theory. The pecking order of capital structure derives from the asymmetric information between managers and investors. According to Myers and Majluf (1984), managers tend to issue new shares when prices are over- valued, thus benefiting old shareholders. Aware of this possibility, new shareholders might demand a discount on the stock price in order to acquire it. Thus, managers avoid issuing new shares, even though this decision can make firms ignore profitable investments. Myers (1984), therefore, suggests that companies seeking to re- duce the costs of asymmetric information have a preference of funding resources. In this sense, companies would prefer using re- tained earnings in first place, then low-risk debt, high-risk debt and, as the last resource, new equity. Hence, companies that have good investment opportunities but lack internal cash flow could turn to debt to fund their projects first, thereby affording such companies high leverage. In contrast, Autore and Kovacs (2010) ng & Finance 35 (2011) 358–371 359 show that firms may issue new equity even in conditions of high asymmetric information, since such asymmetry is lower than the recent past. anki A fundamental difference between the assumptions of the agency theory and the pecking order theory may partially explain the contrasting predictions regarding the influence of growth opportunities on leverage. The agency theory assumes that manag- ers behave opportunistically and rationally, trying to maximize their own utility at the shareholder expense. Leverage, in this case, would discipline their behavior, making companies with few investment opportunities and high free cash flow to increase the use of debt. On the other hand, pecking order implicitly assumes that managers are rational, though not necessarily opportunistic. Thus, in the maturity phase, debt would not have the same disci- plinary effect as the agency theory predicts. In this controversial context, we test whether the relationship between growth oppor- tunities and leverage is positive or negative. A positive relationship would confirm the pecking order theory and a negative relation- ship would confirm the agency theory. In our paper, our proxy for growth opportunities is the ratio of the firm total market value (i.e., debt plus equity market value) to total assets. There is also no consensus regarding the influence of profitabil- ity on capital structure. The pecking order theory, mainly based on the work of Myers (1984) and Myers and Majluf (1984), states a hierarchy of preference in the choice of funding sources, which is defined by the level of information asymmetry. In this context, external equity would be the last resort because of its high level of asymmetric information; debt would come in second, and re- tained earnings would be the first choice. In this sense, Titman and Wessels (1988) suggest that firm profitability is an important capital structure determinant since it reflects the amount of earn- ings that may be possible to firm retain. Thus, Fama and French (2002) suggest that in a simple pecking order model, by holding the investment level fixed, leverage would correlate negatively with profitability. Debt will grow as investment needs is higher than retained earnings. While profitability is frequently treated as a capital structure determinant, Shyam-Sunder and Myers (1999) propose a more direct approach to test the pecking order and also corroborate the theory, contrarily to the studies that show evidence that pecking order does not hold (e.g., Frank and Goyal, 2003; Leary and Roberts, 2010). The trade-off hypothesis, in turn, states a positive relationship because low profitability may increase bankruptcy risk (Fama and French, 2002), thus forcing firms in such a position to adjust their leverage to lower levels. Besides, profitable firms should be more levered as they would benefit from corporate debt tax shields (Frank and Goyal, 2003; Wu and Yue, 2009) in addition to improv- ing the firm performance (Margaritis and Psillaki, 2010) due to the disciplinary role of debt (Jensen, 1986). Again, this controversial context leads us to test whether the relationship between profit- ability and leverage is positive or negative. A positive relationship would confirm the trade-off theory and a negative relationship would confirm the pecking order. In our paper, profitability is de- fined as the ratio of operating income to total assets. For the same reason, the trade-off hypothesis also predicts a negative relationship between distance from bankruptcy and leverage. Thus, financially healthy companies (i.e., with low bank- ruptcy likelihood) tend to have smaller levels of debt. Corroborat- ing this hypothesis, Byoun (2008) find evidence that the larger the Altman Z score (used as proxy for distance of the bankruptcy), the smaller the firm leverage. Hence, our hypothesis is that the longer the distance from bankruptcy, the lower the leverage. Our proxy for distance from bankruptcy is the Altman Z score modified by MacKie-Mason (1990) and is given by Z = 3.3(earnings before inter- est and taxes/total assets) + 1.0(sales/total assets) + 1.4(retained earnings/total assets) + 1.2(working capital/total assets). 360 E.K. Kayo, H. Kimura / Journal of B The firm size is also a very common determinant in capital structure studies. Titman and Wessels (1988) state that larger firms may be more diversified, thereby making them less prone to bankruptcy risk. Also as a function of size, larger firms may have a greater debt capacity. Furthermore, larger companies, being in general more transparent, tend to have larger debt levels and can issue larger amounts of debt, thus allowing them to spread the issuing costs (Byoun, 2008). However, Rajan and Zingales (1995) suggest that this relationship could also be negative. They say asymmetric information problems are likely to be smaller in larger companies. Thus, it would be possible for larger companies to issue new shares (i.e. reducing leverage) with no reduction in the market value. Again, by testing the relationship between firm size and leverage we have two possible results supported by different the- oretical perspectives. A positive relationship indicates the impor- tance of diversification and the opposite relationship supports the role of information asymmetry. In our paper, we use the loga- rithm of sales as a proxy for size. Finally, tangibility plays an important role on capital structure, as the collateral aspects of assets in place tend to increase leverage. In this way, we test the hypothesis of a positive relationship be- tween tangibility and leverage. As Titman and Wessels (1988) sug- gest, since tangible assets can be used as collateral for a given debt, the borrower is forced to use the resources in a pre-determined project, thus curtailing the incentive to assume high risks. Almeida and Campello (2007) show that tangibility is particularly impor- tant when the firm is financially constrained and thus has re- stricted access to external resources. However, according to the results of Almeida and Campello, tangibility is less important when firms are unconstrained. In our paper, tangibility is defined as the ratio of fixed assets to total assets. 2.2. Industry-level determinants Studies on capital structure often employ dummy variables to control the effect of industry on leverage. Nevertheless, few studies analyze determinants of leverage that characterize rather than classify each industry. A rare exception is the study of Simerly and Li (2000), in the strategy field. It is important to remember that one of the theoretical streams in strategy emphasizes the impor- tance of external factors to the firm when determining corporate strategies so as the environmental characteristics similarly affect all organizations of a given industry (Simerly and Li, 2000). In this context, it would be reasonable to suppose that specific character- istics of a given industry could also influence the firm capital struc- ture. However, even Simerly and Li (2000) do not analyze the direct influence of an industry characteristic on leverage. They analyze the influence of leverage moderated by the environmental dyna- mism of the industry on firm performance. Environmental dyna- mism, as suggested by Dess and Beard (1984), reflects the degree