Research in International Business and Finance 24 (2010) 295–314 Contents lists available at ScienceDirect Research in International Business and Finance j o ur na l ho me pa ge : w w w . e l s e v i e r . c o m / l o c a t e / r i b a f Capital structure in an emerging stock market: The case of India Indrani Chakraborty ∗ Institute of Development Studies Kolkata, 1, Reformatory Street, Kolkata, West Bengal 700027, India a r t i c l e i n f o a b s t r a c t This paper applies two alternative methods of estimation, viz., fully modiﬁed OLS (FMOLS) and generalized method of moments (GMM), to analyse the determinants of the capital structure of Indian ﬁrms using a panel of 1169 non-ﬁnancial ﬁrms listed in either the Bombay Stock Exchange or the National Stock Exchange over the period 1995–2008. The results thus obtained are robust across the estimation methods. Among the three alternative theories of capital structure, the pecking order theory and the static trade-off theory both seem to explain Indian ﬁrms’ decisions. However, there is little evidence to support the agency cost theory. © 2010 Elsevier B.V. All rights reserved. Article history: Received 9 November 2009 Received in revised form 4 February 2010 Accepted 4 February 2010 Available online 11 February 2010 Keywords: Capital structure Financial sector Panel data Fully modiﬁed OLS estimation GMM estimation India 1. Introduction Studies on capital structures of corporations have a long history, dating back to the nineteen ﬁfties with the appearance of the works of Lintner (1956), Hirshleifer (1958) and Modigliani and Miller (1958). Theoretical and empirical studies that followed subsequently form an extremely large body of literature.1 Modigliani and Miller (1958) showed that in the perfect ﬁnancial market, under certain assumptions, the value of a company is independent of its ﬁnancing choice. The well-known Modigliani–Miller Theorem is based on several assumptions: in a perfect capital market insiders and outsiders have symmetric information; no transaction cost or bankruptcy cost exists; equity and debt choice becomes irrelevant; and internal and external funds can be perfectly substituted. These assumptions later came under scrutiny and alternative theories emerged which suggested that capital ∗ Tel.: +91 33 2448 8178; fax: +91 33 2448 1364. E-mail address: firstname.lastname@example.org. 1 For an extensive review of literature on capital structure, see Harris and Raviv (1991). 0275-5319/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.ribaf.2010.02.001
296 I. Chakraborty / Research in International Business and Finance 24 (2010) 295–314 structure might be relevant to the ﬁrm’s value. The three main theories that came up subsequently are the static trade-off theory, the pecking order theory and the agency cost theory. In the static trade-off theory (also referred to as the tax based theory) a ﬁrm is viewed as setting a target debt to equity ratio and gradually moving towards it (Myers, 1984). In other words, this theory assumes that some form of optimal capital structure exists that can maximize the ﬁrm value while simultaneously minimize external claims to the cash ﬂow stream. Such claims include bankruptcy cost, agency costs between shareholders and bondholders, and taxes. Thus a ﬁrm’s target leverage is determined by the trade-off between interest tax shields of debt and the cost of ﬁnancial distress. The pecking order theory (also referred to as the information asymmetry theory), developed by Myers and Majluf (1984) and Myers (1984), argues that ﬁrms choose to ﬁnance new investment, ﬁrst by internal retained earnings, then by debt, and ﬁnally by equity. There is no concept of target capital structure for a ﬁrm in the pecking order theory. The explanation provided by Myers for the pecking order theory is based on the assumption that ﬁrm insiders have more information than outsiders. The agency cost theory (Jensen and Meckling, 1976) proposes that the optimal capital structure is determined by agency costs, which include the costs for both debt and equity issue. The costs related to equity issue may include: (a) the monitoring expenses of the shareholders (b) the bonding expenses of the managers and (c) ‘residual loss’ due to the divergence of managers’ decision from those of the shareholder’s (Jensen and Meckling, 1976). On the other hand, debt issue increases the shareholders’ and managers’ incentives to invest in high-risk projects that yield high returns to the shareholders but increase the likelihood of failure that the bond holders have to share if it is realized. If debt-holders anticipate this, a high premium would be charged, which in turn would increase the cost of debt. Thus both equity and debt incur agency costs, and hence the optimal capital structure involves a trade-off between the two types of costs. The empirical studies on the capital structure choices of ﬁrms that started appearing in the eighties (Marsh, 1982; Jalilavand and Harris, 1984; Titman and Wessels, 1988) and continued later are mostly based on data from developed countries. For example, Rajan and Zingales (1995) use data from G7 countries, Bevan and Danbolt (2002) use data from the U.K. and Gaud et al. (2005) analysed data from Swiss companies. There have been a few studies that focus on developing countries as well. For example, Booth et al. (2001) considered data from ten developing countries (Brazil, Mexico, India, South Korea, Jordan, Malaysia, Pakistan, Thailand, Turkey and Zimbabwe), Chen (2004) uses data from China, Pandey (2001) analysed the data from Malaysia, and Wiwattanakantang (1999) uses data from Thailand, and so on. It may be noted here that the institutional structures of corporate ﬁrms of these developing countries are signiﬁcantly different from that of the developed countries. Some methodological issues could be raised in this context. Most of these studies are based on panel data, and they use either a static model or a dynamic model, which simultaneously take care of the heterogeneity of ﬁrms and control for time effects. The models have been estimated by some of the following methods: ﬁxed effects, random effects, pooled OLS and generalized method of moments (GMM). These methods correct for simultaneity bias using instrumental variables and control for unobserved ﬁrm-speciﬁc effects. However, they ignore the integration properties of the data. Therefore, it is not clear from these studies whether they estimate a long-run equilibrium relationship between leverage and its determinants or a spurious relationship which may lead to wrong conclusions. In this paper we apply some recently developed econometric techniques, viz., panel cointegration and panel estimation by fully modiﬁed OLS method, which correct for the shortcomings mentioned above, to provide better insights into the capital structure of non-ﬁnancial ﬁrms in India in the postliberalization period. The issue of capital structure has become very important in India, especially following the gradual initiation of the reform measures in the ﬁnancial sector of India since July 1991. Financing choices of ﬁrms in India remained quite constrained till 1992. Access to the equity market was controlled by the Controller of Capital Issues which imposed severe restrictions on ﬁrms (Bhaduri, 2000). In May 1992, the Controller of Capital Issues was abolished and ﬁrms were allowed more freedom of access to the equity market. In 1994 the National Stock Exchange (NSE) was set up with nationwide stock trading and electronic display and clearing and settlement facilities. Due to the competitive pressure from the NSE, the Bombay Stock Exchange (BSE), the oldest stock exchange in India, also introduced electronic trading in 1995. Certain reform measures were initiated in the banking sector at the same time which enhanced the choice of ﬁnancing by ﬁrms through debt too.
I. Chakraborty / Research in International Business and Finance 24 (2010) 295–314 297 These reform measures include, ﬁrst, the deregulation of interest rates by the banking sector. Second, some liberalization measures have been taken on the cash reserve ratio (CRR) and statutory liquidity ratio (SLR). Before 1991, the CRR was as high as 25 per cent and SLR was 40 per cent. The CRR has come down to 5 per cent and SLR is 24 per cent at present. Third, since 1991, a number of foreign banks and private entrepreneurs have been invited to commence banking operation in India. The numbers of foreign and private banks operating in India increased from 21 and 23 in 1991 to 33 and 30 in 2004, respectively. Finally, in March 1996, uniform prudential norm was introduced in the lines of Basel Committee on Banking Supervision. Very few banks had a capital adequacy ratio of up to 8 per cent before 1991. By March 1998 only one of the 28 public sector banks fell short of this standard (Ahluwalia, 1999). Following the reform measures there were efforts to reduce the nonperforming assets (NPA) too, which came down to 1.3 per cent by the end of 2007–2008 (Government of India, 2009). As a result of these reform measures in the ﬁnancial sector of India the capital structures of Indian ﬁrms have changed signiﬁcantly. This provides an opportunity to study the changing nature of ﬁnancing decision of Indian ﬁrms. The issues such as how the listed non-ﬁnancial Indian ﬁrms ﬁnance their projects, and after liberalization what the determinants of these ﬁrms’ capital structure are, have been analysed in great detail in this study. The analysis is conducted using a balanced panel data pertaining to 1169 Indian non-ﬁnancial ﬁrms for the period 1995–2008. Altogether 16,366 observations have been available for the analysis. Given that the ﬁnancial reforms were initiated in 1991 it is expected that the impact of reforms would be felt only after a couple of initial years. Hence we begin our analysis from 1995. Our results show that both the pecking order theory and static trade-off theory are at work in the capital structure choice decisions of Indian ﬁrms. However, there is little evidence to support the validity of the agency cost theory. This paper is organized as follows. In Section 2, we provide an overview of the literature on the measures of leverage and the determinants of the capital structure. Descriptive statistics on data are provided in Section 3. Section 4 deals with the empirical analysis and we conclude in Section 5. 2. Measures of leverage and the determinants of capital structure The earlier empirical studies used two measures of leverage as dependent variable, viz., book leverage and market leverage. Book leverage is deﬁned as the book value of total debt divided by the book value of total assets. Market leverage is deﬁned as the book value of total debt divided by the book value of total liabilities plus the market value of total equity. Since there is no information on the market value of total equity in the database on which this study is based we take only the book value in our measures of leverage. We use two measures of leverage in this study, viz., the ratio of total borrowing to asset (LEV1) and the ratio of total liability to sum total of total liability and equity (LEV2). Equity is considered at 365 days average closing price. All the variables are measured in terms of the book value. The ﬁrst of these two measures was used in an earlier study on Indian ﬁrms by Bhaduri (2002) and the second measure was used by Huang and Song (2006). Among the determinants of capital structure, we discuss proﬁtability, tangibility, size, growth opportunities, non-debt tax shields, and uniqueness. 2.1. Proﬁtability The theoretical prediction about the effect of proﬁtability on leverage is ambiguous. According to the pecking order theory, ﬁrms use internal sources of ﬁnancing ﬁrst and then go for external sources of ﬁnancing. Firms with higher proﬁtability will prefer internal ﬁnancing to debt and hence a negative relationship is expected between proﬁtability and leverage. Most empirical studies conﬁrm the pecking order hypothesis (Titman and Wessels, 1988; Rajan and Zingales, 1995; Michaelas et al., 1999; Booth et al., 2001; Chen, 2004). According to the static trade-off theory, more proﬁtable ﬁrms are supposed to have more debt-serving capacity and more taxable income to shield. Therefore, according to this theory, when ﬁrms are proﬁtable they are likely to prefer debt to other sources in order to beneﬁt from the tax shield. Hence a positive relationship is expected between proﬁtability and leverage. We consider two alternative measures of proﬁtability. In the ﬁrst measure, we consider
298 I. Chakraborty / Research in International Business and Finance 24 (2010) 295–314 proﬁtability as the ratio of proﬁt before interest, tax and depreciation to total assets (PROFT1). In the second measure, we consider proﬁtability as the ratio of cash ﬂows to total assets (PROFT2), in the line of Bhaduri (2002). 2.2. Tangibility According to the agency cost theory, there are incentives for shareholders to invest in a sub-optimal manner due to conﬂicts between lenders and shareholders. Because of this tendency, lenders will take actions to protect themselves by requiring tangible assets as collateral. Firms with high levels of tangible assets will be in a position to provide collateral for debts. If the ﬁrm defaults on debt, the tangible assets will be seized but the ﬁrm will avoid bankruptcy. It is therefore expected that a positive relationship exists between tangibility and leverage. Some studies from the developed countries report a signiﬁcant positive relationship between tangibility and total debt (Titman and Wessels, 1988; Rajan and Zingales, 1995 among others). However, the ﬁndings from the developing countries are mixed. Wiwattanakantang (1999) observes a positive relationship between tangibility and leverage in Thailand but Booth et al. (2001) for ten developing countries and Huang and Song (2006) for China ﬁnd a negative relationship. Following Huang and Song (2006) and Bevan and Danbolt (2002) we measure tangibility as the ratio between ﬁxed assets and total assets (TANGY). 2.3. Firm size The effect of ﬁrm size on leverage is ambiguous. Rajan and Zingales (1995) argue that larger ﬁrms generally disclose more information to outsiders than smaller ones. Larger ﬁrms with less asymmetric information problems should tend to have more equity than debt and hence have lower leverage. Therefore, following the pecking order theory of capital structure, it is expected that the size of the ﬁrm would be negatively related to leverage. On the other hand, according to the trade-off theory, larger ﬁrms tend to be more diversiﬁed and thus less prone to bankruptcy. This argument suggests that ﬁrm size should be positively related to leverage. A large number of studies ﬁnd positive relationship between ﬁrm sizes and leverage (Booth et al., 2001; Wiwattanakantang, 1999; Huang and Song, 2006; Rajan and Zingales, 1995 among others). On the other hand, Bevan and Danbolt (2002) observe that ﬁrm size is negatively related to short-term debt and positively related to long-term debt. We use natural logarithm of sales as a proxy for the ﬁrm size (SIZE). 2.4. Growth opportunities Firms with higher growth opportunities would need more fund. According to the pecking order theory, there will be stronger preference for external ﬁnancing, especially for debt. Hence we expect a positive relationship between growth and leverage. On the other hand, as discussed earlier, ﬁrms with growth opportunities may invest sub-optimally and therefore creditors will be more reluctant to lend for longer periods (Myers, 1977). In such a situation the problem can be solved by short-term ﬁnancing or by convertible bonds (Titman and Wessels, 1988). Therefore, we expect short-term debt to be positively related to growth if growing ﬁrms go for short-term ﬁnancing instead of long-term ﬁnancing. Rajan and Zingales (1995) and Booth et al. (2001) ﬁnd positive relationship between growth and leverage. In our study we use two alternative measures of growth opportunities. Following Titman and Wessels (1988) we take the percentage change in total assets (GRTH1) as our ﬁrst measure. Our second measure of growth opportunities is the percentage change in sales over the year (GRTH2), following Chen et al. (1999). 2.5. Non-debt tax shields Firms are likely to favour debt because they can beneﬁt from the tax shield due to interest deductibility. Thus we expect a positive relationship between effective tax rate and leverage. However, DeAngelo and Masulis (1980) argue that non-debt tax shields (such as tax deductions for depreciation and investment tax credits) are substitutes for the tax beneﬁts of debt ﬁnancing and a ﬁrm with larger
I. Chakraborty / Research in International Business and Finance 24 (2010) 295–314 299 non-debt tax shields is expected to use less debt. Therefore an increase in non-debt tax shield can affect leverage negatively. This relationship is corroborated empirically by Wald (1999), Chaplinsky and Niehaus (1993) and Huang and Song (2006). Following Huang and Song (2006) we use the ratio of depreciation and amortization to total assets as the measure of non-debt tax shields (NDTS) in this study. 2.6. Uniqueness Titman (1984) argues that a ﬁrm’s capital structure should depend on the uniqueness of its product. If a ﬁrm offers unique products, its customers, workers and suppliers suffer relatively high costs in case of liquidation and hence the costs of bankruptcy increase. Accordingly, the trade-off theory predicts a negative relationship between uniqueness and leverage. We use research and development expenditures over sales as the measure of uniqueness (UNIQUE). 3. Data We use annual data on Indian non-ﬁnancial ﬁrms, for the period 1995–2008, listed either in the Bombay Stock Exchange or in the National Stock Exchange, as available from the Centre for Monitoring Indian Economy’s database PROWESS. The sample is a balanced panel on 1169 ﬁrms for which a continuous data set exists over the sample period. All ﬁrms with any missing observations for any variable during the sample period have been dropped. In aggregate, we have 16,366 observations. Firms which operate in the ﬁnancial sector (banks, insurance companies and investment trusts) are not included in this analysis since their balance sheets have a different structure from those of the non-ﬁnancial ﬁrms. The summary statistics of the major variables for selected years (1995, 1999, 2004, and 2008) as well as for the entire period 1995–2008 are presented in Table 1. We get similar pictures for the trends of leverage from the two alternative measures. Both the measures of leverage, LEV1 (measured by the ratio between total borrowings to asset) and LEV2 (measured by the ratio between total liability and the sum total of total liability and equity) increase over the period whereas in 2008 they turn to the values almost similar to 1995. However, the two measures of leverage differ sharply both in individual years as well as during the entire period 1995–2008. During the entire period 1995–2008, LEV1 is 0.355 whereas LEV2 is 0.758. It is evident that over the period the ﬁrms have grown in size and the share of tangible assets in their balance sheets has remained all the same. The variable UNIQUE, measured by the research and development expenditure over sales, also exhibits a slightly rising trend over the years. Table 2 reports the correlation coefﬁcients between the variables. The two alternative measures of leverage are not correlated, as the correlation coefﬁcient is 0.083. Among the explanatory variables, non-debt tax shield is highly correlated with tangibility (correlation coefﬁcient is 0.389) and the two alternative measures of proﬁtability, PROFT1 and PROFT2, are also highly correlated (correlation coefﬁcient is 0.411). Moreover, size is highly correlated with one measure of leverage, LEV2 (correlation coefﬁcient is 0.319). To check if these high correlation coefﬁcients between non-debt tax shields (NDTS) and tangibility (TANGY), two measures of proﬁtability and between ﬁrm size (SIZE) and LEV2 would create serious problem of multicollinearity, we conducted the test for variance inﬂation factor (VIF). VIF tests reveal that the value corresponding to each explanatory variable is much less than 10, which indicates that multicollinearity is not a serious problem here.2 Thus the inclusion of NDTS and TANGY in the same model speciﬁcation should not create any problem. We already stated that we measured equity at 365 days average closing price. Following Quah (1993) we make an attempt to construct a “mobility matrix” for equity of the 1169 ﬁrms over the 14year periods from 1995 to 2008 to understand the changing pattern of equity relative to the average of these 1169 ﬁrms. To construct the mobility matrix ﬁrst we take the ratio of equity of each ﬁrm to the average equity of all ﬁrms in 1995 and 2008. If for any particular ﬁrm this ratio is less than 2 Multicollinearity is a serious problem if the value of the variance inﬂation factor (VIF) is greater than 10 (Nachane, 2006).
300 I. Chakraborty / Research in International Business and Finance 24 (2010) 295–314 Table 1 Summary statistics for leverage and its determinants. Variables 1995 Mean LEV1 LEV2 PROFT1 PROFT2 TANGY SIZE GRTH1 GRTH2 NDTS UNIQUE 0.326 0.627 0.145 0.043 0.388 3.731 73.909 209.716 0.022 0.002 Std. dev. 0.232 0.292 0.083 0.493 0.193 1.759 208.352 1431.100 0.017 0.008 1999 Mean 0.347 0.817 0.107 0.085 0.424 4.176 6.477 13.713 0.036 0.008 Std. dev. 0.265 0.212 0.138 0.140 0.200 1.822 19.098 97.875 0.024 0.003 2004 Mean 0.373 0.817 0.117 0.075 0.383 4.443 9.598 18.316 0.036 0.008 Std. dev. 0.855 0.200 0.162 0.103 0.198 2.049 53.543 74.825 0.023 0.005 2008 Mean 0.346 0.697 0.138 0.067 0.356 4.881 20.071 53.669 0.031 0.003 Std. dev. 0.503 0.257 0.326 0.235 0.208 2.371 40.073 743.736 0.027 0.018 1995–2008 Mean 0.355 0.758 0.124 0.080 0.396 4.327 17.664 37.408 0.034 0.003 Std. dev. 0.639 0.243 0.187 0.203 0.201 2.008 71.155 475.649 0.026 0.027 Min. 0 0.002 −5.590 −1.833 0 −4.605 −93.137 −99.997 −0.023 0 Max. 51.777 1 9.933 16.527 0.996 12.508 3113.636 31,850 0.637 1.942