POLICY RESEARCH WORKING PAPER 320 Var,ia'bl''eD' s 'm prdk cap al Capital Structures h predict sucure in the United Sttes in Developing Countries also precict choicesof ap . structnfe in a sample of tn . s , _ . ~~~~~~~~~~~- d~~eveloping::aUraies. in: Evidence from Ten Countries e i n seve -rcountries. tota. indebtednes is negatvely Aslh Deniqga4-Kunt relatedto, net fixed assets. Voiislav M&smovic--:---- -- Vsuggesting tw markets for long4erm debt do not function efitaely. The World Bank Policy Rcseach Deartment Fmnance andi Pivate Sectr Development Divsion July~ ~ ~~~~~~~~~~~~~~~A 199 ------ PoLicy RESEARCH WORKNG PAPER 1320 Summary findings Demirgfiq-Kunt and Maksimovic investigate capital Still, variables that predict capital structure in the structures in a sample of the largest publicly traded firms United States also predict choices of capital structure in in ten developing countries - Brazil, India, Jordan, the the countries sampled. Republic of Korea, Malaysia, Mexico, Pakistan, Variables suggested by agency theory explain more of Thailand, Turkey, and Zimbabwe - for 1980-91. the variation than variables suggested by tax-based The firms in the sample are smaller than comparable theories. For both short-term and long-tern equations in U.S. finns, and the financial systems and regulations in most countries, the asset structure, liquidity, and industry thcse countries differ significanly from those in the effects have more explanatory power than firm size, United States. Not every country has well-functioning growth opportunities, and tax effects liquid financial markets in which investors can diversify In several countries, total indebtedness is negatively risks. Nor do all countries have efficient legal systems in related to net fixed assets, suggesting that markets for which a broad range of property rights can be enforced. long-term debt do not function effectively. This paper-a productof the Financeand Private Sector Development Division, Policy Research Department-is part of alarger effort in the department to study emerging stock markets. The sntuy was funded by the Bank's Research Support Budgct under the research project "Stock Market Development and Financial Intermediary Growth" (RPO 678-37). Copies of the paper are available free from the World Bank 1818 H Street NW, Washington, DC 20433. Please contact Bill Moore, room N9-055, cxtension 35261 (38 pages). July 1994. The Poliy Reearch Workig Paper s f lisdemnas the fhdig of work h proas to excowu the exdcang of ideas about deelopment issues An obecte of dte seris is to get the findi Got qucy, ven if thepresoewatiouare kss than fidly polishe The papers cary the names of the asors ad shodd be usd ad cited accardingly. The findings, intepreations, and concusions are he auth'r own and sould not be attried to the Wordd Bank its Excuie Board of Diors or any of its menber counrie Produced by the Policy Research Dissemination Center Capital Stnrctures in Developing Countries: Evidence from Ten Couwy Cases Ash Demirgui-Kunt Vojislav Maksimovic* *Thc authors are at the Policy Research Department, The World Bank and the University of Maryland, College Park, respectively. We thank Herb Baer, Jerry Caprio, Stijn Claessens, and Ross Levine for helpful comments. We are indebted to Sara Zervos for excellent research assistance and Bill Moore for help with the manuscript. . .Introduction There exists a large literature that describes and attempts to explain observed capital structure choices in developed economies (see the survey by Harris and Raviv (1990). This literature normally takes as given, first, the existence of well fimctioning liquid financial markets in which investors can diversify risks and second, the existence of an efficient legal system in which a broad range of property rights can be enforced. While appropriate for the analysis of financing choices in the U.S., these assumptions are often not satisfied in other economies. In many economies financial markets are at early stages of development: only a small proportion of the risks is traded, the markets are relatively illiquid and heavily regulated. Little is known about the effect of such conditions on the opfimal financing choices of firms. In this paper we provide evidence on the capital structure choices of firms for a panel of ten developing countries using annual data for 1980-1991. We describe observed financial structure choices and compare them to U.S. financial structure choices. We ask whether models of developed for United States institutions explain capital stucture choices in our panel of developing countries. While r.ar findings bear most directly on developing economies, they are of more general interest. Models oFfinancial structure have been developed with the aim of explaining U.S. data. They are based on U.S. institutions. Predictions often depend on specific assumptions about institutional structure of these models. Testing the models with data from economies with less developed financial markets and very different institutions provides a test of robustness of these models. In this sense our paper can be viewed as a complement to the work of Bradley, Jarrell 3 and Kim (1984) and Titman and Wessels (1988) for the U.S., and Ragan and Zingales (1994) for a sample of several developed countries. The the paper is organized as follows. Section II discusses the principal differences between the U.S. and the developing economies that we are analyzing. Section HII discusses the theoretical framework. Section IV presents the empirical results and Section V concludes. IOL Insttutions in Developing Countries We investigate capital structure in Korea, Malaysia, Brazil, Mexico, Turkey, Jordan, Zimbabwe, India, Thailand and Pakistan. We identify four broad reasons why developing country capital structures may be different: differences in the level of economic and financial development, institutional dlifferences, smaller firm size, and different tax treatment of debt and equity. A. Economic and Financial Develovment: As shown in Table 1, the per capita GDP in these countries is much lower tha in the U.S. It ranges from about 22.5% of the U.S. level in the case of Korea to justl1.9% in the case of Pakdstan. These economies and their financial systems. differ in other significant respects from that in the U.S. In this section we review some of these differences and discuss how they may affect the firmn's capital structure decision. One measure of market development is the ratio of market capitalization to Gross Domestic Product, MCAP/GDP. As revealed in column (3) of Table 1, MCAP/GDP ranges from 16. 1% in the case of Turkey to 70% in the case of Malaysia, with a median of 23 %. The corresponding statistic for the U.S. is 74.5%. However, MCAP/GDP of most of the countries 4 in our sample is greater those in some developed countries, such as Germany and Italy (Pagano (1993)). Thus, stock markets in most of the countries in the sample are consequential in their economies. Financial intenmediaries are another source of financing for corporatioDns. While it is difficult to measure the development of finncial internediaries, it is likely that the availability of financing is positively related to the size of financial intermnediary sector. Hence, the size of the formal financial intermediary sector relative to economic activity has frequently been used as a measure the finanial sector development.1 Column (2) of Table 1 lists the ratio of M3, the liquid liabilities of the financial system, to GDP.' The Table reveals a wide variation iti our sample. It reflects Jordan's role as a regional financial center and the relative sophistication of Malaysia's financial sector. Corporations' ability to rais external financing may be related to the stability ofT the price level. Mean inflation f:ates over the sample period are listed in column (5). Our sample contain countries with very high rates of inflation (Brazil, Mexico and Turkey) and very low rates (Jordan and Malaysia). As expected, the inflation rate is negatively correlated with the ratio of M3 to GDp.3 B. Institutional Differences: There are some significant institational differences between the financial system in our sample of countries and those in the Uwited States. In capital markets, certai pricing decisions '- See, for example. Goldsmith (1969) and McKinnon (1973). 2 Liquid liabilities arc mecasured by M3. which is defined as the currency held outside the banking system plus demand and interest bearing liabilities of banks and nonbank fmiancial intermediaries. 3 The Spearman rank coefficient is -0f97. 5 are much frequently more heavily regulated, whereas there are often fewer protections for investors. Table 2 summarizes some key institutional features of the markets in our sample. A key deviation from U.S. practice is that with the exception of Brazil, Mexico and Jordan, in all the countries in our sample there exist restrictions on the pricing or issue of bonds or stocks. In some cases these restrictions are significant, particularly for new issues. For example, during our sample period in Pakistan companies could not offer shares at above book value (Mirza (1993, page 208)). Similar restrictions existed in India (Glen and Pinto (1994)). In Malaysia, "looming over the entire industry is the feared Central Issues Committee" which, among other powers, has the right to determine the prices of initial public offerings and which may take up to six months to rule on a price (Seaward (1993), page 153)." As revealed in Table 2, fimancial markets in our sample countries differ in the amount of protection offered to investors. According to data compiled by the Intematonal Finance Corporation, accounting standards are adequate or of internationally acceptable quality for all countries except Jordan.' By the end of 1992 all the countries in the sample had a securities commission or a similar- government agency. India, Korea, Malaysia Mexico and Zimbabwe also have bond rating agencies. Differences from U.S. practice are not confined to the regulation of capital markets. Governments of the countries in our sample are in general more active in business affairs. Most importntly, in some countries financial intermediaries are required to provide directed long-tenn 4Se Seward for an argumet tha thCapital Issucs Commitess pices of initial public offerings artificially low (page 152). Ritter (1993) provides evidence that average underpricing of initial public offerings in Malaysia is 149%. 5 in some cases some requirements appear to be stricter han in the U.S In Malaysia the Central Issues Committee holds directors of newly floated companies rcsponsible for deviations of morc than 10% frem forecasts in prelisting prospectuses (Seaward (1993)). 6 credit to selected firms (see, for example Baer et al (1994) and their references). C. Differences in Firm Size: A further difference between the U.S. market and the countries studied here is in the size of firms. As shown in Table 3, publicly traded firms in the U.S. are much larger than publicly traded firms in our sample countries. Only in Korea does the average size of firms in the highest quartile exceed the average size of firms in the lowest quartile in the U.S. At the other extreme, in Zimbabwe, the average firm in the largest quartile is six times smaller than the average firm in the lowest U.S. quartile. However, although these firms are small by U.S. standards, they are large relative to their local economies. Hence, they may receive attention from local financial communuities that U.S. firms of comparable size would not. D. Tax Treatment of Debt vs. Eguity: Table 4 summarizes the tax treatment of interest income, dividends and capital gains. In most countries the personal income tax rates vary with income. To provide a consistent comparison of tax levels across countries we have assumed that the marginal investor is a private individual who is sufficiently wealthy to be paying personal income taxes at the highest rates.6 Using this benchmark, interest payments are clearly the most tax advantaged form of payment in only two countries: India and Korea. In Brazil, Malaysia, Mexico, Pakistan and Thailand the net tax burden is generally lower on equity income. m. Detemiinants of Capital Structure 6 Rajan and Zingales (1993) make the same assumption in their comparison of tax burdens in a sample of developed economies. Similarly, when there exists multiple corporate tax rates we have assumed that firms face the highest raze. Thresholds for the maximum rate are typically set very low. 7 In this section we fintroduce a framework for discussing the firmn's choice of financial structure in developing economies. We review the existing literature on financial structure and discuss its applicability to developing economies. The corporate finance literature has focused on two broad determinants of capital structuire: agency theoretic explanations that stress conflicts of interest between various stakeholders in the firm and explanations that stress tax consequences of capital structure choices. We review each in turn and describe the variables that we use to measure the predicted effects. A. Agencv Teoretic Exuplanations of Capital Structure: Severl distinct conflicts of interest that arise between the investors holding different classes of securities have been identified in the literature. These conflicts arise because holders of one class of investors (typically equityholders) act as agents for other investors and take decisions that affect the value of the firm as a whole.' As a result, these investors have an incentive to engage in opportunistic behavior that increases their payoffs at the expense of other classes of investors and the firm as a whole. The reduction in the firms value that results from opportuistic behavior by those in -ontrol of a corporation is termed the agency cost of fiancing. If they are rational, the holders of securities whose value is reduced by opportunistic behavior factor their expected losses into the price that they are willing to pay for their securities. Hence, it is in the finns' residual owners' interest to choose capital structe that minimizes agency costs, and thereby maximizes the price at which each fiam's securiies can be sold. As a result of these choices, if agency theoretic explanations are valid, the observed capital structure of each firm should depend on '7 For example. die equiiyholders may make decisions chat alter the risldness of the firm's operations. 8 the potential for opportunistic behavior in that firm. The potential for opportunistic behavior depends on the extent to which the agents' actions affect value and the extent to which contracts that regulate actions can be written and enforced. Thus, agency costs will depend on the firm's technology, development of financial institutions and markets, the investors' incentives to monitor and the legal system in each country, among other factors. The firm's optimal capital stnxcture will in general also depend on these factors. The corporate fmance literature has identified several conflicts of interest that arise in many situations and analyzed fmancial structures that minimize them. The two most important conflicts are between the firm's insiders and outside investors and the conflicts between equityholders and debtholders. The firms insiders frequently have opportnities to consume perquisites in ways that cannot be easily monitored by outside investors. This creates a conflict of interest between insiders and outside investors. This conflict can be mitigated in following ways: * By issuing debt securities instead of equity the insiders can contractually commit themselves to a prespecified level of payment to outside investors, thus reducing opportunities for opporaunistic behavior. * By issuing debt securities with shorter matuities the insiders commit themselves to renegotiating the firm's financing at short intervals. This reduces their insiders to exploit their creditors. A second important conflict of interest is that between the firm's equity holders and the 9 firm's debt holders.' As leverage increases, the equityholders have an incentive to siphon funds out of the firm through dividends and stock repurchases. This is because all the siphoned funds go to the equityholders whereas the consequent reduction in the finn value is shared with the debtholders. In addition, because the value of equity is a convex function of the. value of the firm, as leverage increases equityholders have an incentive to select risky policies even if they lead to decreases in firm value. The conflict between equityholders and debt holders can be reduced by: * Reducing debt levels in industries where the potential for opportunistic behavior is high. * Securing long term investments with specific capital assets. * Shortening the maturity of debt. As the maturity of debt is a critical variable in agency models, we analyze the firm's choice of long-term and short-term debt levels separately. Our measure of long term indebtedness is the ratio of the book values of long term debt to total equity, LTDTE, and our measure of short-term indebtedness, STDTE, is the rtio of the book value of short-term debt to the book value of debt. The determinants of financial structure are:9 Asset Structure: The composition of the firm's assets affects its abilitry to commit not to engage in opportunistic behavior. Fixed assets usually have collateral value. A finm with fixed assets can issue secured debt, thereby limiting its ability to expropriate the debt-holders. Thus, we expect firms with greater amounts of fixed assets to issue more long-term debt than finns with fewer fixed assets. We use the ratio of net fixed assets to total assets, NFATA as our e For a detailed analysis see Hart (1993). 9 We give below the interpretations of variables that we find most plausible. Additional interpretations of some variables are possible, as in for example Tiunan and Wessels (1988). 10 .' lmeasure of the firm's asset structure. We expect that this ratio is positively related to LTDTE and negatively related to STDTE. In the absence of collateral, a greater degree of monitoring by creditors may be optimal. Monitoring by creditors is facilitated by issuing debt with shorter maturity. Thus, we would expect firms which do not borrow to finance fixed assets to have more short tenn debt. A measure of the firn's financing needs, other than the need to finance fixed assets, is the Ratio of Net Sales to Net Fixed Assets, NSNFA. Firms with a high ratio of sales to net fixed assets have cost structures requiring more monitoring, and are therefore expected to have more short- term debt and less long-term debt. Leverage and Liquidity Constraints: Recent literature, following Myers-Majluf (1984) has suggested that internal generated capital is cheaper for the firm than external financing. This suggests that firms would finance internally first, and issue debt only when such low cost sources of fmancing have been exhausted. We measure the finn's initial excess internal funds by DIVCSH, the ratio of dividends paid out to shareholders to its cash flow available for reinvestment: earnings after taxes plus depreciation. The higher this variable the less cash constrained the firm is. Thus, we would expect both LTDTE and STDTE to be negatively related to DIVCSH.10 An alternative variable that has been used to measure the firm's ability to generate capital internally, by Titman and Wessels (1988) among others, is earnings before interest and taxes over total assets, PROFIT. In studies of the U.S. economy this variable has repeatedly been 10 For some countries data on depreciation and eamnings after tax is not available. For hose countries we use the ratio of dividends to total assets. DIVTA, to proxy for cash constraints. 11 found to be negatively correlated with leverage and we expect it to be similarly related to financing choices in our sample. Growh: As suggested by Myers (1977), equityholders in highly leveraged firms with significant growth opportunities have incentives to adopt suboptimal investment policies. If this agency cost is significant, we expect fast gro-wing firms to be financed with equity or with short- tern debt. Our measure of growth is the rate of growth of real total assets, GROWTH. We expect this variable to be negatively related to LTDTE and to be unrelated or positively related to STDTE. Firm Size: Thme is considerable evidence that firm financing patterns in developed countries differ for firms according to size. Barclay and Smith (1993) have shown that size is an important predictor of debt maturity in the U.S. and Baias and Hillion (1991) have shown that the amount of short term credit is predicted by firm size. This effect may arise because access to financial markets may be a function of firm size. Additionally, the amount of monitoring by investors may depend on the liquidity of the market for the firm's equity, which, in tum, is related to firm size-" We allow for these effects by segmenting our sample into quartiles by size of total assets and icluding size dummies in our equations, SZ1-SZ4, where SZl denotes the smallest quartile. An alternative size vaiable we also use is firm total assets to GDP, TAGDP. Firm Age: There are arguments supporting the prediction that younger firms will be less indebted (Titman and Wessels (1988)). Younger firms have shorter credit histories which makes it more difficult to judge their quality. Younger firms also tend to be riskier since they exit As shown by Riuer (1987), there exist economies of scale in the issuance of equity. 12 more frequently. To test for thie age effect we cnst-~c twaaibe:teaeo h im AGE, and a dummny variable, YOUNG, which t-ake the value one if AGE is less tha or equal to five, and zero otherwise. Indu~stry Classification: The pkoduct market structur and type of competitive interaction across fi-rms differs from industry to industry. To the extent that capital structur affect the incentives of firms to enter into imnplicit contracts with rivals (Maksimovic (1988)) or to maintain reputations (Maksimovic and Titman (1990)), capital structures will differ systematically across industries. To control for these industry effects, we include when available dummies for inLdustry classification (at the 2 digit SIC code level)."2 These industry dummies may also pick up differences in asset structure across industries that are not captured by NFATA and NSNFA. Market 's Valuation of Equity: Our measures of the firm's capital structur, the ratios of the book values of long-term and short-term debt to total equity, implicitly assume that book values adequately measure the economic values used in determining the firm's capital structure. However, book values do not directly measure the market's valuation of the firm's growth opportunities. If the finn can borrow against the value of growth opportunities, firim's with higher markcet valuations will have higher book value of debt to equity ratios (LTDTE and STDTE).13 To contol for this we include the difference between the market's valuation of the firm's equity and its book value, scaled by the book value, (MV-BV)IBV, as an explanatory variable. We expect this variable to be positively related to LTDTE and STDTE. B. Taxation and the Capital Structure: Firms were classified into 2-digit SIC codes on the basis of descriptions of the firm's principal industry. 13 itis unalky thatdefirmn borrowagai its growth opporne in thesame way thatit an borrow against fixed asse. For a discussion of these issues see Myers (197 and Hat and Moore (1991). 13 The second imnportant determinant of capital structure is the tax system. The fin-n's financing choice affects. its tax liabilities because the total amount of taxes paid by the firm's investors, at both the corporate and personal levels, differs accordinig to whether they bold equity or debt securities. This differential treatment of investment income induces investors' preferences for holding equity or debot securities. Firms attempt to satisfy these preferences by optimailly altering their capital structur. Depending on the level of personal and corporate taxes, this optimizing behavior by finns results in one of two outcomes. First, if tax rates are such that one formn of financing (debt or equity) is unambiguously tax advantaged, then there may exist an optimal debt-equity ratio for each firmn that minimizes its total taxes and that depends on the firm's tax liabilities." Second, for some tax regimes the aggregate supply of debt and equity securities in the economy may adjust so that individual firms are indhfferent between issuing debt or equtity (Miller (1977h). In both cases individual firms may have an incentive to chose low debt levels if they cannot utilize debt tax shields (DeAngelo and Mastilis (1980)). We examine the relationship between an individual firm's debt level and two measures of non-debt tax shields, depreciation expanse over total assets, DEPTA, and estimated non-debt tax shields over total assets, NDTS. Following Titman and Wessels (1988), we estimate the latter as NDTS = EBT - (EBT-EAT)/Tc where EBT is income before taxes, EAT is income after taxes and Tc is the corporate tax rate. 14 This follows from the analysis of (Modigliani-Miller (1958)). 14 IV. Empirical Results A. Discussion of the Data: The data for this study were collected by the International Finance Corporation. The description of the data set and the definition of each variable are given in the Appendix." The means of the variables used in the study are shown in Table 5. Inspection of Table 5 reveals that in every case for which we have data, the mean short- term debt exceeds the mean long-tf.rm debt. India, Korea and Pakistan have the highest mean levels of total debt and Brazil, Mexico and Zimbabwe have the lowest levels. These contrasts are also shown by time series of aggregate levels of short-term, long-term and total debt by country in Figure 1. B. Regression Results: In this subsection we examine the finnial capital structures of the firms in our sample. We follow the literature, in regressing measures of capital structure STDTE and LTDTE on the detenninants discussed above.16 In interpreting the results of the regressions it is necessary to keep in mind that short-term debt and long-term debt are both components of total indebtedness. In some cases theory predicts different determinants of long-term and short-term debt 25 We have deleted some observations from the data set when they seemed to contain obvious errors or to pertain to situations not within the scope of this research. Thus, for example. observations which did not report fill fiscal year results were deleted. Likewise, observations which reported implausibly highly negative growth rates were also eliniinated- Details arc available from the authors. 16 This is also consistent with the theoretical models on which the above list of capital structure determinants is based. These models derive capital structure as a function oF given determinants. This suggests staistical tests in which the determinants are taken as exogenous. A more general approach would be to derive a strucural model of financing in developing economies. Our approach may be viewed as an initial step towards that approach. 15 indebtedness: asset structure, for example, is predicted to affect each differently. In other cases, only the total level of debt may matter: for example in shielding income from taxes. In the latter case, the sign of any single component of debt may differ from that predicted by theory for the total level of debt.'7 Consider first asset structure as a determinant of capital structure. Agency theory suggests that firms with large fixed assets have a comparative advantage in obtaining long-term debt, whereas finns with high sales relative to fix assets have a comparative advantage in borrowing over shorter periods. This suggests that NFATA is positively related to long-term indebtedness LTDTE and that NSNFA is positively related to short-term indebtedness, STDTE. By implication, we expect that firms NFATA is negatively related to STDTE and that NSNFA is negatively related to LTDTE. Panels (a) and (b) of Table 6 reveals that of the eight countries for which we have data on these variables (we do not have data on net fixed assets for Zimbabwe) the relationship between NFATA and LTDTE is positive and significant at the 5% level or better in five of the eight cases. The relationship between NFATA and STDTE is even more striking. In seven of the eight countries the relationship is negative and significant- The only exception is Mexico, where it is positive but not significant. The signs of the coefficients of NSNFA also support the predictions, although they are less clear cut. NSNFA is significant at the 5 % level or better and positively related to STDTE in four cases. It is negatively related in only two cases, and they are not statistically significant. " We have rerun the regressions reported below, normalizing debt with total assets instead of total equity. The general conclusions are unaffected. 16 Taken together, these results offer strong support for the prediction that finms with large amounts of fixed assets have a comparative advantage in borrowing over the long-term. They offer more qualified support for the proposition that firms with high ratios of sales over net fixed assets have a comparative advantage in short tenn borrowing. The finding that finns with large net fixed assets have a comparative advantage in long- term borrowing does not imply that such firms have a comparative advantage in long-term borrowing over equity financing. Panel 6(c) shows the results of the regression of total debt on total equity, TDTE. NFATA in the TDTE equation is significant and negative in four cases and significant and positive in one case, that of Mexico. NSNFA is significant and negative in two cases. It is also significant and positive in two other cases. These results imply that, if anything, firms with more fixed assets have less total debt than fims with fewer fixed assets. This finding is counterintLutive in a partial equilibrium framework that abstracts from the state of development of the market for credit- However, it may be explained as discussed by Shleifer and Vishny (1992) and Worthington (1994), if asset specificity reduces the coliateral value of fixed assets, or if the market for long-term debt financing in the sample countries is less developed than the market for equity fmancing.'8 Next, consider the effect of the firms cash flow on its capital stmcture. Our preferred measure of excess funds, DIVCSH is available for five of the nine countries. Of these five cases it is significantly related to STDTE in three instances. The relationship is negative in all these cases. DIVCSH is significandy related to LTDTE in three instances, all of them negatively so. For four countries we do not have information to calculate DIVCSH. In three of these 3- This would be the case, for example, if properLy rights of long-term debt-holders are cosdy to enforce. 17 cases (Brazil, Jordan and Turkey) we proxy for excess cash flow by using the ratio of dividends to total assets, DIVTA.19 Four of the six coafficients have the predicted negative sign. However, only one, in the STDTE for Turkey is it statistically significant. Inspection of Table 6 reveals an even stronger negative relationship between profitability and leverage. PROFIT is negatively related to STDTE for all nine countries (eight of them significantly) and also negatively related to LTDTE in eight of the nine countries (five of them significantly).2Y The exception is Pakistan, where PROFIT is positively, but insignificantly related to LTDTE. However, even in the case of Pakistan there is a significant negative relationship between PROFIT and STDTE. The signs of the liquidity and profit variables, DIVCSH, DIVTA and PROFIT, in the TDTE equations are also consistent with a negative relationship between profitability and leverage. The signs are significant three, two and eight times respectively and are uniformly negative when significant. Taken together, these results point to a strong negative relationship between internally generated resources and indebtedness - profitable firms and firms that make large payments to equityholders borrow less. In conjunction with the fmding above that finrs with high net fixed assets do not take advantage of their asset structre to borrow more in total, perhaps as a result of monitoring issues identified above, this suggests that a lack of liquidity, may be an important determinant of debt financing in the sample countries. 21 This variable measures the ratio of cash paid out to equityholders to total assets. Thus, it is a less accurate measurc of the firm's voluntary payouts to cquityholders than DIVCSH, which measures the proportion of available cash- flow paid out. 20 For Jordan and Mexico profit is defined as EBT instead of EDIT. Therefore, the nregative relationship is built in. 18 -A' : The relationship between the rate of growth and leverage is weak but consistent with predictions of agency theory, in particular Myers (1977). The theory predicts that fast growing firms will issue short term debt in preference to long term debt in order to align the incentives of equityholders and bondholders. For six of the nine countries in our sample the coefficient of GROWTH in the equation explaining STDTE is positive. It is only statistically significant in two of these cases. The coefficient is not statistically significant in either of the three cases in which it is negativc. In the equation explaining LTDTE the pattern is even weaker. Only one of the four positive coefficients is significant. One of the five negative coefficients is also significant. The results for the TDTE equation are equally inconclusive: the coefficient of the growth variable is more mixed: it is positive and significant in two of the four cases and negative and significant in one. The excess of market valuation to book valuation (MV-BV)IBV is predicted to be positively related to STDTE and LTDTE.2' We have market value data for five countries. Inspection of Table 6 shows that for all five the coefficient of (MV-BV)/BV is positive in the STDTE equation. It is statistically significant in three of the five cases. For four out of the five cases the coefficient of (MV-BV)/BV is positive in the LTDTE equation. However, it is only signifcant in two of these cases. The results for the TDTE equation are similar: the coefficient of (MV-BV)/BV is positive in each case, but is only significant in two of these cases. Tax effects are measured by two variables. The most direct measure is the estimate of the firm's non-debt tax shields, NDTS. Theory predicts that debt and non-debt tax shields are 21 More preisely, this variable is predicted to be positively related to dte debt level, not necessarily to each of the components of indebtedness. 19 substitutes. The coefficient of NDTS in the STDTE equation is significant in four instances. It is negative in each case. In the LTDTE equation the coefficient is negative in five of the eight cases, but is significant in only two. Similarly, the coefficient of NDTS is negative in seven out of the eight cases in the TDTE equation. However, it is significant in only three cases, and for Mexico the coefficient is positive. The other measure of non-debt tax shields is DEPTA, depreciation over total assets. Data are available for five countries. Inspection of Table 6 rcveals that the coefficient of DEPTA in the STDTE equation is significant (and negative) in three of the five cases. It is significant and positive for only Malaysia. The relationship between LTDTE and DEPTA is significant only in the case of Kprea, where it is negative. Turning to the size dummies, the coefficient of SZ4 in the LTDTE equation is positive and significapt in five of the nine cases (it is insignificant in the other four cases). The coefficients for SZ2 are predominantly negative and all three of the significant coefficients are negative. This suggests that largest firms find it easier to obtain long term financing compared to smaller firms which find it more difficult. There is no similar pattern in the STDTE equation. When we substitute size dummies with. total asset size to GDP, TAGDP, in the LTDTE equation the coefficients are positive for seven out of nine countries and significant in four of them. The results of the STDTE are mixed. Finally, our age variables do not have significant coefficients in any of our regressions. One reason for this is that the firms in our sample tend to be mostly older, over ten years of age. In our equations we have included industry dummies. This specification is consistent with 20 methodology used in previous cross-sectional studies of capital structure such as Bradley, Jarrell and Kim (1984) and Titman and Wessels (1988) for the U.S. The inclusion of industry dummies is also motivated by the considerable theoretical literature that predicts industry effects.22 In order to investigate the effect of including industry effects we have reestimated the equations reported in Table 6 without the industry dummies and with individual firm dummies replacing the industry dummies. T.he two alternative specifications have differing implications for the interpretation of the regression equations. The specification without the industry dummies risks attributing to the variables included in our study differences in capital structure across industries that are caused by omitted variables. By contrast, if individual firm dummies are included, then the explanatory power of variables de ining firm level results may be reduced because differences in firms' capital structures are "explained" by firm specific dummies. When industry specific dummies were dropped, R2 coefficients fell. However, the nmber of significant coefficients increased.23 By contrast, including firm specific dummis alters the results to a greater extent. Whereas the R2 coefficients increase, some of the explanatory variables lose significance. In a few instances, particularly that of India, Brazil and Turkey more variables become significant. The significant variables retain their signs. 22 Theoretical models by Tian (1984), Maksimovic (1988), Maksimovic and rrm (1990) and others predict industry effects. Empirical evidence on industry effects is provided by Chevalier (1993) and Phillips (1994)- 23 For example the GROWTH coefficientwhich had been insignificant become positive and significant in tWo instances in the LTDTE equation (Malaysia and Turkey) and one instace in the STDTE equation (Brazil). The most interestng case was chat of (MV-BV)IBV, which gains significance in the STDTE equation in two instances (India and Malaysia). In the LTDTE equatioD (MV-BV)/BV lost significance in one case aIndia) and gained in another (Zimbabwe). 21 Our sample also includes few, publicly controlle enterprises.2 However, excluding them from the sample does not alter the results significantly. C. Relative Exnlanatorv Power of Different Determinants of Calpital Structure: In this section we analyze the relative importance of different determinants of capital in explainig the variation in debt levels for each country in the sample. We compare the explanatory power of regressions that include all the variables identified above as determinants of financia capital structur with regressions in which some variables are deleted. The difference in explanatory power provides a measure of the importance of the deleted variables in explainig sample variation. Table 7 presents the decomposition of RI for short-term and long-term debt in each country in the sample. In accordance with the discussion above, the dtrinants of capital structure are classified into asset structure, liqidty, growth opportunity, size effect, tax effect and industry effect variables. Inspection of Table 7 reveals that the rankings of the avenage explanatory power of all the variables are very similar for short-term and long-term debt equations (last three rows), the only difference being in the relatve importance of industry effects and asset structure. In both of the debt equations the industry effects and the asset structur composition have the greatest explanatory power, followed by liquidity effects, the size effect, growth opportunities and taxes.2 This is consistent withi an inspection of individual country results, in which asset 24 One in Korea,threin TUrkey,nand onein Mexico. 2X if Brazil is omitted from short-term averages idustry effects still have the grtest explaatory power. however thqUisiy effect becomes moe imporn th ran asset satucture. 22 structur and liquidity variables, together with industry effects usually have more explanatory power tha growth opportuiWties, size effects and tax effects. All the variables with the exception of the tax and size effects have on average more explanatory power in the short-term than in the lor.g-tcnno debt equations. One interpretation of these results is that in our sample markets for short-term financing may be functioning better tha markets for long term capital. V. Concusions In this paper we have tested agency theoretic and tax-based explanations of capital structure choice on data from a panel of developing countries. In these countries financial markets are underdeveloped and much more heavily regulated than in the U.S. The fimns in the sample are much smaler than U.S. fims on which financial structure models are usually tested. Tax treatment of debt and equity also vary considerably in each country. Despite these differences, both the agency theoretic and tax-based models of capital structure predicted capital stuctures in our sample well. Net fixed assets are positively related to long-term debt and negatively related to long-term debt. More profitable fims and firms that are making large payouts to shareholders have less debt. Firms with high market to book ratios have more debt. Finns with high non-debt tax shields have less debt. interestingly, total indebtedness is negatively related to the ratio of net fixed assets to total assets. Thus, even fims with assets that could serve as collateral, finance themselves by retned earnings or equity issues rather dt by issuance of long-term debt. This suggests that 23 markets for long-term credit do not function effectively in several countries in the sample. A comparison of the relative explanatory power of the determinants of capital structure shows that for both short-term and long-term equations in most countries the asset structure, liquidity and industry effects have more explanatory power tha firm size, growth opportunities and tax effects. The explanatory power of theory in our sample is quite strong when compared to Titman and Wessels' (1988) study using US data. This may be in part due to differences in market efficiency and the legal systems. If alternative contractal means of resolving agency conflicts are not avaialable, then the relationships derived from simple agency-theoretic models are more likely to have empircal validity. In this paper we focused on fim level differences in capital structure within each country. In our fuiture work we will investigate cross-country differences in capital structure in greater depth. 24 References Baer, H.L., Caprio, G., Demirguc-Kunt, A. and D. Vittas, 1994, Termn finance: Theory and Evidence, World Bank mimeo. DeAngelo, H. and R. Masulis, 1980, Optimal capital structure under corporate and personal Laxation, Journal of Financial Economics 8, 3-29. Diamond, D., 1991, Debt maturity structure and maturity risk, Quarterly Journal of Economics 106, 709-737. Glen, J. and B. Pinto, 1994, Debt or equity? How developing country firms choose, IFC mimeo. Harris, M. and A. Raviv, 1991, The theory of capital structure, Journal of Finance 46, 297-356. Hart, 0., 1993, Theories of optimal capital structure: a managerial perspective, in Blair M. (ed.), The Deal Decade, (The Brookings Institution, Washington DC) Hart, 0. and J. Moore, 1989, Default and renegotiation, a dynamic model of debt, MIT Department of Economics Working Paper No. 520. Bradley, M., G.A. Jarrell and E.H. Kim, 1984. On the existence of an optimal capital structure: theory and evidence, Journal of Finance 39, 857-877. Maksimovic, V., 1988, Capital structure in repeated oligopolies, Rand Journal of Economidcs 19, 389-408. Maksimovic, V. and S. Titman, 1990, Financial policy and reputation for product quality, Review of Financial Studies 4, 175-200. Mirza, K., 1993, Pakistan: A small market with potential, in The World's Emerging Stock Markets, edited by Keith K. H. Park and Antoine W. Van Agtmael, Purbus Publisiing Co., Chicago, Illinois. Myeras, S., 1977, The determinants of corporate borowing, Journal of Financial Economics, 147-76. Myers, S., and N. Majluf, 1984, Corporate fmiancing and investment decisions when firms have information that investors do not have, Joural of Finanaial Economics, 187-221. 25 Pagano, M., 1993, The flotation of companies on the stock market, European Economic Review, 1101-1125. Park, K. and A. Van Agtmail, 1993, The World's Emerging Stock Markets, (Probus, Chicago IL) Rajan, R. and L. Zingales, 1994, Is there an optimal capital structure? Evidence from international data, University of Chicago mimeo. Seaward, N., 1993, Malaysia, in The World's Emerging Stock Markets, edited by Keith K. H. Park and Antoine W. Van Agtmnael, Probus Publishing Co., Chicago, Illinois. Shleifer, A. and R. W. Vishny, 1992, Liquidation values and debt capacity; A market equilibrium approach, Journal of Finance, 47 (4), pp. 1343-1366. Titman S. and R. Wessels, 1988, The determinants of capital structure choice, Jounal of Finance, 1-21. Worthington P., 1994, Investmcnt, cash flow, and sunk costs, Federal Reserve Bank of Chicago, mimeo. 26 LTDTrE LTDrTE LTDITE - __D CO m t:> _ cm C." D 10 -0 CD ~~~~~~~~~~ C2 CD -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~- LE I TDtTE ID - -~ E - L m rn~ - ,,, ,I _ r _ C _ CD~~~~~~~~~~~C - , - mX~~- Kl- ~ ~ ~ 0 LTnrrFE LTD/TE LTD/TE > co CD ~~~~~~~~CD - CD ~~~~~~~~~~o; 0~~~~~ _ _ | _ H m- m t2 LOTELT1E TDT I .U *~ I _~~~~~~ _ CD CD.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ >- 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ' STDITE STD1TE STD.JTE o a' c o o' _ - o 0.~~~~~~ Co . o_ o . a'- h- - ,~~~~~~~~~~~~~~C _ m co- 10 >S , :>eCO >~ c ~~~-u a a m ~ ~ ~_ _ __ >10~ ~ ~ . -~~~~~~ STDrrE STD/TE STDITE e- , o va e - CD . ' -n 5 10. * t G cm cm CC) r-j-~~~~~~~~0~ c~~~~ _ _ M_ m- CD C- M~~ ~ cm STDITE STDITE ST D/TE o u. 0 01 o cn 10. _D aCD - C) , cm CD - 10C. M1C co C.LI~~~~~~~~~C 0- 0- 0- u.m -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C m ~~ ZI :1- m z m' 10~~~~~~~~~~~~~~~ > tD-> cm > g° _ UZ~~~ 0 *w 10 * m -~~~~~~~~0- o ; ~~~~~~~~o o LO ~~~~~~~~~~~~~~~~~~~~10C m _ - ^ : 0-~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 0-; *~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 0..~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~- ..0 Figure l.c. TD/TE NO .io_n D 141 * 9950 1982 10Z4 1986 luau I84 g0100 I02 1109t hI'Ji 10 IUW 980 92 IOU 1 98 199 YEAR YEAn YEAR ... .. .. . . 5 ..... .... . . . . 5 ,MEX 5 IiLI 112 911 11). 013 919 0 Iil 01f~911 110 11W 914 , IV IIJU 1ll lll lll0 I--luau 1-0l52 ll-I 4ll l LllI lEri-- 0--1- 1942 I4 iuv IDWl§lllI YEAR YEAA YE58 t A R 9 S PAK TIIA. IJR °- 10|||||lit-1-o- ODO0 9982 998 9988 1go 990 19 132 4 9095i 91138 9900 9980 9982 9989 9966 19998 1999 YEAR YEAn TEAR ZIMt -. 980 9952 998 19aa 1965 199 Table 1. Economic and Financial Development Indicators - 1991 GDP/CAP is the GDP per capita in US$. M3/GDP is the currency held outside the banking system plus demand and intcrest bearing liabilities of bznks and nonbank financial intermediaries divided by GDP. MCAP is the stack market capitalization in millions of US$. MCAP/GDP is the stock market capitalization divided by GDP. Average annaul inflation is given for the period 1980-1991. GDPe CAP M3JGDP MCAP (tie$) MCAP/GDP Avemage Annual ___________ ($) Inflation 80-91 (%) United States 18,934 67.0 4,180,210 74.5 4.2 Korea 4,268 52.3 96,373 38.4 5.6 Malaysia 2.449 114.67' 58,627 70.0' 1.7 Brazil 1,893 13.1 42,759 28.3 327.6 Mexico 1,812 25.0 98,178 22.8 66.5 Turkey 1,375 29.4 15,703 16.1 44.7 Jordan 1,372 135.9 2.512 55.3 1.6 Thailand 1,304 71.4 35,815 39.4 3.7 Zimbabwe 653 42.7 1.394 36.0 12.5 India 369 44.2 47,730 17.5 8.2 Pakistan 359 36.5 7.326 19.1 7.0 1988 Figure 27 Table 2. Developing Countries -- Institutional Factors Column (1) 0=published, I =comprehensive and published internationally; Columns (2) and (3), 0-poor, I =adequate, 2=good, of internationally acceptable quality; Column (4), 1 -functioning securities commission or similar government agency, 0=no agency; Column (5), O=free, I =some restrictions, 2=restricted; Column (6), O=no restrictions, I -restrictions; Column (7) 1 =formal rating agency, 0-no formal agency. All data are as of 1992. Columns 1-5.are based on the information provided in the EmerainE Markets Factbook published by the International Finance Company. Columns 6-7 are from The World's Emerging Stock Markets, 1993, by K. Park and A. Van Agtmail, published by Porobus, Chicago and sources In the World Bank. Country (1) (2) (3) (4) _ (5) (6) (7) Regular Accounting Investor Securities Restrictions on: Restrictions Formal Publication of Standards Protection Commission on pricing or Rating P/E yield i issue of bonds Agency Idividend capital foreign or stocks repat. repat. entry Brazil 1 2 2 I 0 0 0 0 0 Mexico 1 2 2 1 0 0 0 0 I India 1 2 2 I 1 1 1 I Korea 1 2 2 1 0 0 1 1 1 Malaysia 1 2 2 1 0 0 0 1 1 Pakistan 0 1 1 1 0 0 0 1 0 Thailand I _ I 1 0 0 0 I 0 Jordan_ _ _ _ __ 0 0 0 0 0 Turkey I I ° 1 0 0 0 1 0 Zimbabwe 0 1 I 2 2 11 Indian restrictions were abolished in May 1992. 28 Table 3. Average Size in Each Quartile (in US$) The values are average $ total assets, for each quartile of firms classified by total assets, over the country's sample period. SALL AIEDIUM LARGE VERY LARGE US 4,350,000 26,030,000 147,680,000 7,909,830,000 KOREA 630,000 1,210,000 1,780,000 5,270,000 MALAYSIA 198,000 401,000 760,000 3,403,000 INDIA 283,000 572,000 898,000 2,860,000 MEXICO 59,000 180,000 443,000 2,106,000 JORDAN 41,000 96,000 173,000 1,778,000 BRAZIL 99,000 178,000 308,000 939,000 THAILAND 50,000 106,000 200,000 850,000 TURKEY 78,000 176,000 292,000 814,000 PAKISTAN 57,000 118,000 176,000 765,000 ZIMBABWE 59,000 116,000 210,000 644,000 29 .u. Table 4. Tax Advaniiage of Debt with rcspect to Dividend and Capital Gains The fix rates used arc the slalulory onies. Data are obtained from various editions of Coopers & Lybrand, InlemalionalTax Summaries. COUN1fY KOREA lUAYSIA INDIA . NEXICO BRA7.L TIURKEY PAKISTAN Z IM-AUWKr THAnAND YEAR 1910 1990 1911 1990 1980 1990 1914 1990 19U4 1l9 19e2 1 I990 1980 19185 190 1983 19S2 1990 XI corporate tax rate 0420 0.3175 0.500 0.390 0.391 0.52i 0.420 0.360 0.450 0.450 0.400 0.492 0.525 0.400 0.495 0.500 0.300 0.300 Xl corporte tax rate on distributed profltS 0.420 0.375 0.500 0.390 0.591 0.55 0.00 0 000 0.450 0.450 0.400 0.492 0.525 1.400 0.495 0.500 0.300 0 300 XI high pcracnal tax tate 0.744 0.600 0.350 0.400 0.720 0.325 0.530 04.0 0.600 0.300 0.630 0.500 0.660 0.450 0.495 0.600 0.630 01i50 X, pOfaufal capitIz gains u; 0.744 .000 0.000 0.000 0.720 0.52S 0.510 0 450 0.600 0.250 0.650 0.500 0.000 0.0W 0.000 0.300 0.000 0.000 Xs mate on interst income 0.744 0.600 0.550 0.400 0.720 0.525 0.550 0.450 0.600 0.500 0.650 0.50a 0.660 0.450 0.495 0.600 0.650 0.550 X4 rate an dividend Income 0.744 0.600 0.400 0.350 0.720 0.525 0.oo 0.000 0.230 0.010 0.850 0.S00 0.660 0.450 0.200 0.200 0.650 0.5S0 XI tax tel ate en dIvidendr 0O.50 0.120 0.40D 0.550 0.000 0.000 0.000 0. 000 0.000 0.00 0.330 0 330 0.000 0.000 0.000 0000 0.350 0. 300 X. net intetcll income per Sl 0.255 0.400 0.450 0.600 0.230 0.475 0.410 0.5510 .400 0.500 0.350 0.500 0.340 0.550 0.505 0.400 0.350 0.*50 X, net capital salns per SI 0.141 0.250 0.500 0.610 0.114 0.226 0.261 0*352 0.230 0.413 0.210 0.254 0.475 0.600 0.505 0.350O 0.70W 0.70 X, el dividendspef SI 0.235 01S 0.500 0.610 0.114 0.226 1.00a I W 0.414 0.506 0.405 0.421 0.162 0.330 0.404 0.400 0.490 0.525 XI1taediaa2vanialeofdividenoJaw. r.t. Jeh 0.0S0 0111 *0.111 *0.017 0.591 0.515 -1.222 .0.116 0.D59 *0.012 *0.166 0.157 0.525 0.400 0.200 0.000 .0.400 -0.167 X31laxdindvantageadcspitalgainsw.r.tdclbt 0.420 0.375 .0.111 *0.017 |0591 0.52$ 0.420 0.360 0.450 0.17S 0.400 0.492 .0.397 40.091 0.0o 0.125 -1.000 4.5*6 XII - I - X,. X - (I *-X, (I(4). X's - (- X XJ) (I.X + XI). XC - (X, - X5)/XI krid XI, - (XI - XJYX. 30 able S. Capital S4ructure - Descriplive Satistics by Country LTD\TE ii hb book value of lonSg ter debt divided by book valu of equity. STDITE and TDITE& re the book value of short itan and otal debt divided by book value or equity. NFATA is the nct fixed aaseu divided by lotl ssets. DIVCSH Is the dividends divided by eamings afer Laxes plus deprecialion. DIVTA Is the dividerds divided by total ssets. GROWTH Is the growth te oo teal total assets. PROFIT is the incone beforc interest end taxcs divided by lotal ssets. (MV*RV)IBV is the market vAlue of equity minus book value of equity divided by book value of equity. NSNFA Is the net sales divided by net fixed *as.t NDTS is the nond :bt tas shield which Is earnings before Uxes minus thc tatlo of corporate taxes paid to corporate taic rate, dclated by otalI assets. TA/GDP Is total Assets divided by the GDP of the eountry. Te value of each item is calculated as dtt average of all lirms for tach country's sample period. Extended variable definitions and sources are given in the Appendix. LTDITE uTrnrr TDrTE NFATA DEPTA DIVCSII DIVTA GROWFH PRtOFIT MtV-DY NSNFA NDTS TAIGDP ERAZII. .139 .421 .560 .640 .002 .170 .057 - 1.164 517 - friA .745 1i350 259S .411 .039 .193 .01e .109 .132 .135 5.691 023 .0006 JORDAN .274 MM 1.180 .469 . .032 .077 .065 .439 2 760 oOS7 KOREA 1.05E 2.391 3.449 .371 .053 .122 OD18 .104 .100 -.21S 4 336 002 .0023 MALAYSIA .278 .833 1.111 .465 .023 .350 .024 .170 .035 1.24? 3.025 003 .004t MEXICO .401 .417 .818 .569 .034 .079 - 1.463 .013 - PAKISTAN .596 2.359 2.955 .334 .03E .232 102 .010 .115 11.144 055 .0011 lURKEY .4S5 1.509 1.994 .415 - 068 .080 .238 4 231 010 .0010 ZINIMADWE .187 .615 .802 .031 .260 .028 .034 .131 -.437 033 0062 THIAIILAND = 2.332 .398 = .041 .223 2.021 5.310 0007 31 Table 6.a. Capital Structure In Developing Countries -LTD/TE Estimated model is: LTDITR - e + p, SZ + I, IN + aJT + 0, NPATA + Os DEPTA + 5. DIVCSH (or DIVTA) + 5, GROWTH + 0, PROFrT + 0, (MV-BVIBV) + P, NSNFA + 0, NDTS e. te dependent vaiiable Is long term debt to cquit ratlo. Definitions of Independent variables are given ins ihe appendix. Regression is estimated usinB Ordinary Least Squars. Not repaited below are secgor(IN) ad year (T) dummy variablcs. White's heseroakedasticity-corrmced standard errors are giYCn in parenthesls. and * indicate dial ihe coetficilent i signinleanly different frm seto at I and 5 pereent levels respectely. KOREA MALAYSIA INDIA MEXICO JORDAN BRAZIL TURKEY PAKISTAN Z[MBABRWE NFATA 1,498" -.310 1.704" 6.146" .617" -.112 .297 1.992" (.409) (.176) (.233) (2.247) ( 195) (.060) (.402) (.462) DEFrA -7.382" -.916 -1.301 -2.023 -.068 (1.192) (1.526) (1.097) (1.327) (.787) DIVCSH -.735 .033 -1.355" -.547" -.27t" (.528) (.027) (.564) (.154) (.086) DIVTA .389 -.607 -1.175 _____________ (.872) (.696) (,782) GROWfTH -.881" .137 -.054 -.002 .051 -.008 .296 .667" -.050 (.247) (.088) (.125) (.003) (.109) (.005) (.184) (.264) (.075) PROFIT -5.607" -2.5 15" -.435 -3.389" -.181 -.445" - 1.176" .464 -.466 (1.386) (.532) (.345) (1.158) (.329) (.1 18) (.276) (.554) (.473) MV IY .832' .025 .026' -.029 .173 BV (.421) (.014) (.012) (.052) (.124) NSNFA .002 -.011I' .002 .545" -.006 -.002 -.028- -.002 (.010) (.003) (.002) (.209) (.009) (.008) (.012) (.002) NDTS -4.263" .402 -I.11I 3.420 -.268 .142 -2.427" .136 (1.807) (.250) (.841) (1.583) (.216) (.294) (.646) (.586) SZ2 -.163' -.058 -.203" .163 -.088 -.023 -.117 .171 -.121" (.082) (.044) (.064) (.173) (.082) (.019) (.081) (.157) (.043) SZ3 -.020 .187" -.046 - .050 .039 -.003 -.138 -.205 -.083" ____________ (.070) (.057) (.073) (.289) (.079) (.022) (.081) (.107) (.041) SZ4 .688" .181" -.089 .907' .241 .095 -.064 -.126 .033 (.141) (.065) (.066) (.434) (.062) (.026) (.108) (.137) (.059) Rs| .26 .20 .30 .30 .35 .24 .23 .28 .21 adj R' .23 .15 .27 .21 .28 .22 .14 .25 .17 N 982 630 743 277 320 598 . 325 809 321 32 Table 6.b. Capital Structure in Developing Countries - STD/TE Esirmatcd mulel is: STDITE - a + p SZ + B, IN + B,T + Bd Nr'ATA + B, DEPrA + 0, DIVCSII (or DIVTA) 4 p, GROWT}I + p, PROFIT + p, (MV.DVJBV) + 0,, NSfFA + pi, ND7S + * Te dependent varlabic It shon crmn dcbt (o equicy ratio. Derinilions of Independent varablcs ate given in te. appetdix. Regression Is estinuted using Ordinaiy Leust Squares. "ot reported btelow ar setor N) ad year (T) dummy variablts. White's heecroskedasticity-correcled standard errors are given In parenthesis. and * Indicate thal the coerrcient is significnly ditferenA from zero at I arid peret leves respectively. . - - - KOREA MIALAYSIA INDIA MEXICO JORDAN BRAZIL TURKEY PAKISTAN ZIMBABWE NFATA -1.733" -.732" -1.339" .819 -1.606" -1.096" -2.652" -2.833" (.567) (. I 10) (.385) (.527) (.418) (.140) (.505) (.820) DEPTA .4.914" 8.532" -7.576" -3.086 -1.9S54 (2.001) (2.509) (1.820) (2.715) (.547) DIVCSII -1.056 -.031 -1.914" -1.928- -.164- L____ _ (.808) (.043) (.919) _ (.553) (.078) DIVTA -1.324 1.711 -2.917" . __________ ____________ _____ _ ___________ _(1,956) (2.477) (1.088) GROWThI .289 -.002 -.149 -.001 .223 .009 .629" 1.341" .104 (.370) (.047) (.238) (.001) (.181) (.010) (.155) (.514) (.079) PROFIT -4.090" -2.670" -4.493" -2,098" -3.135" -.519 -3.240" -3.923" -.940 (Z.021) (.625) (.706) (.420) (1.124) (.285) (.623) (.931) (.380) MV-BV .542" .076 .052 .305"' .404" BV (.214) (.074) (.035) (.146) 0 _ (.3) NSNFA .003 -.003 .035" . 11S' .023 .051" .007 .018' (.009) (.002) (.010) (.061) (.026) (.019) (.024) (.004) NDTS -9.184' -.918' .2.270 .541 -.972" -.821 -5.678" -.501 (4.458) (.433) (1.780) (.509) (.420) (.441) (1.908) (.378) SZ2 -.089 -.140" -.510" -. 10d .109 -.023 .111 1.024' .033 (.139) (.065) (.127) (.091) (.225) (.038) (.189) (.584) (.050) SZ3 -.176 237' -.175 -.347" -.115 -.008 -.027 -.491 -.166" .____ _ (. 160U) (.097) (.114) (.128) (.204) (.040) (.135) (.432) (.046) SZ4 .465" .097 -.440" -.079 .338 .132" .665" -.381 -.177" _ _ (.166) (.081) (.111) (.122) (.194) (.056) (.216) (.418) (.054) R__ ._ '31 .29 .37 .28 .46 .43 .43 .19 .24 adJ R' .29 .25 .34 .19 .40 .42 .37 .14 .20 N 981 630 743 277 320 s 598 325 809 321 33 Table 6.c. Capital Structure in Developing Countries -TDITE Eslimaled model is: TDITB = a + fit SZ + f13 IN + 13T + 13 NPATA + 1j DEPTA + p, DIVCSH (or DIVTA) + PI GROWTH + a, PROFIT + P, (MV.BV/B"j + ga. NSNFA + p1, NDTS + e. The dependeni variable is total debt to equity ratio. Delinitions of independent variables are given In the appendix. Regressilon is estimated using Ordinary Least Squarr.. Not mrponed below are sector (IN) and year (T) dummy variables. White's heirmskedasticity-corrected slandard errors ard given in parenihesis.* and ' indicate that the coerficient is signricantly different from zer at I and 5 percent levels respectively. - - _ . ________________ |_______ KOREA MALAYSIA INDIA MEXICO JORDAN BRAZIL TURKEY PAKISTAN ZINMBARWE 1THAILAND NFATA -.555 -1.042" .365 6.965" - .998' -1.20B" -2.355" -.856 -.247 5.7S9) (.206) (.543) (2.611) (.478) (.159) (.669) (1.074) (.894) DEPrA -10.S05" 7.616" -8.878" -5.119 -2.022" (2.781) (2.779) (2.635) (3.722) (1.023) DIVCSH -1.845 .002 -3.270". -2.475" -.435' (1.347) (.054) (1.431) (.674) (.146) DIVTA -.935 1.105 -4.092' -8.678" _____________ (2.177) (2.24B) (1.531) _ _ (3.411) GROWFH -.541 .135 -.204 - .003 .273 -.001 .925" 2.009" .053 -.699" (.483) (.093) (.331) (.004) (.240) (.013) (.291) (.671) (.127) (.261) PROFIT -10.252" -5.185" -4.928" -5.4864' -3.316' -.974" -4.417" -3.461" -1.406 (2.758) (.863) (.974) (1.379) (1.185) (.311) (.729) (1.114) (.752) MV-BV 1.341- .101 .078 .276 .577" .153 DV (.619) (.074) (.045) (.170) (.221) (.089) NSNFA .008 -.014" .037" .663" -.028 .049" -.022 -.017"* .008 ___________ (.015) (.004) (.012) (.241) (.032) (.022) (.029) (.005) (.026) NDTS -14.278"' -.516 -3.384 3.961' -1.240" -.680 -8.107" -.364 (5.973) (.512) (2.481) (1.828) (.545) (.493) (1.489) (.787) SZ2 .095 .083 -.713" .063 .021 -.047 -.006 1.201 -.087 -.048 (.194) (.083) (.174) (.221) (.248) (.047) (.224) (.686) (.082) (.343) SZ3 -.237 .425" -.220 -.397 -.076 .005 -.165 -.690 -.!49" .508 (.191) (.116) (.166) (.363) (.239) (.052) (.168) (.468) (.079) (.429) SZ4 1.136" .278" -.529" .828 .580" .228" -.601i -.501 -.143 .965s |__________ (.253) (.107) (.159) (.514) (.220) (.068) (.272) (.488) (.102) (.308) R' _.33 .32 .25 .29 .42 .42 .38 .18 .21 .19 adj R2 .31 .29 .22 .20 .35 .40 .32 .15 .17 .12 N 981 630 743 277 320 598 325 809 321 371 34 lable . Delenninanis of Capital Structure - Relative Explanatory Power Reported numbers are Oie adjusted R's after deleting the specified variables. The difference between the adjusled RI of the fill model and ihat of the deleted models are given in parmnhesis and ate avenged bi do . last rows. Variable definitions and sources arc given in the Appendix. Full Asset Structure Liquidity Growth Size Effect Tax Effect Industry Effect Omits NFATA, Omits Opportunities Omits Size Omits NDTS Omits Industry NSNFA DIVCSHIDIVTA Omits Growth Dummies DEPTA Dummies _________ _______ ____________ Profit (MV-BV)/BV . _ KOREA LTD .23 .22 (.01) .20 (.03) .16 (.07) .20 (.03) .21 (.02) .20 (.03) ST) _.29 .28 (.01) .28 (.01) .28 (.01) .28 (.01) .27 (.02) .19 (.10) MALAYSIA LTD .15 .13 (.02) .11 (.04) .13 (.02) .13 (.02) .15 (0) .13 (.02) STD .25 .23 (.02) .23 (.02) .24 (.01) .25 (0) .24 (.01) .17 (.08) INDIA LTD .27 .20 (.07) .23 (.04) .26 (.01) .26 (.01) .26 (.01) .25 (.02) STD .34 .27 (.07) .25 (.09) .34 (0) 33 (.01) .33 (.01) .33 (.01) PAKISTAN LTD .25 .20 (.05) .24 (.01) .24 (.01) .24 (.01) .23 (.02) .18 (.07) STD .14 .12 (.02) .12 (.02) .14 (0) .13 (.01) .13 (.01) .11 (.03) TURKEY LTD .14 .13 (.01) .11 (.03) .13 (.01) .15 (-.01) .15 (-.01) .10 (.04) STD .37 .26 (.11) .24 (.13) .35 (.02) .33 (.04) .36 (.01) .33 (.04) JORDAN LTD .28 .25 (.03) .28 (0) .28 (0) .25 (.03) - .21 (.07) STD :40 .36 (.04) .32 (.08) .37 (.03) .38 (.02) - .30 (.10) MEXICO LTD .21 .11 (.10) .20 (.01) .21 ( 0 ) .19 (.02) .20 (.01) .16 (.05) STD .19 .18 (.01) .14 (.05) .19 (0) .16 (.03) .19 (0 ) .16 (.03) BRAZII, LTD .22 .21 (.01) .21 (.01) .22 (0) .18 (.04) .22 (0) - STD .42 .16 (.26) .42 (0) .42 (0) .41 (.01) .41 (.01) ZIMBABWE LTD .17 - .10 (.07) .14 (.03) .12 (.05) .18 (.01) STD .20 - .17 (.03) .08 (.12) .13 (.07) .18 (.02) - THAILAND TD .12 .12 (0) .09 (.03) .09 (.03) .11 (.01) - .10 (.02) Average LTD (.038) (.027) (.017) (.022) (.008) (.043) STD (.068) (.048) (.021) (.022) (.001) (.056) All . (.049) (.037) (.020) (.022) (.009) (.047) 35 Appendix: Variablc Definitions and Sourccs Variables: LTD/TE Le ng tenn debt to total equity STD/TE, Short term debt to total equity TD/TE Total debt to total equity NFATA = NFA/TA Net fixed assets divided by total assets DEPTA = DEP/TA Depreciation divided by total assets DIVCSH = DIV/(EAT+DEP) Dividends divided by earniings after taxes plus depreciation DIVTA = DIV/TA Dividends divided by total assets GROWTH = ((TA/GDPDEF)-LAG(TA/GDPDEF))/LAG(TAIGDPDEF) Growth rate of real total assets PROFIT = BIT/TA Earnings before interest and taxes divided by total assets. If EBIT is Pot available EBT is used instead (MV-BV)/BBV Market value of equity minus book value of equity, divided by book value NSNFA = NS/NFA Net sales divided by net fixed assets 36 -NDTS = (01 - FE - (EBT-EAf)/CTR)/TA = (EBT - TAX/CTR)/TA Non-debt tax shields equals earnings before taxes (operating income minus financial expenses) minus the ratio of corporate taxes paid to corporate tax rate, all deflated by total assets SZ1-SZ4 Size quartile dummies. SZ1 takes the value 1 if the company is in the first asset quartile in the country and zero if it is not. TAGDP = TA1GDP Total asset size of the company relative to GDP of the country SIC1-SIC61 Sector dummy variables for different SIC codes AGE = YEAR - ESTYEAR Year minus year established. YOUNG Dummy variable that takes the value I if AGE is less than or equal to 5 and zero otherwise. Sources: Firm level variables are constncted from IFC's Corporate Finance data set (see Table Al for a summary). Tax data are from Coopers & Lybrand, International Tax Summaries, John Wiley and Sons, N.Y., various issues. 37 Table Al. IrC's Corporate Finance Data Base |COUNTRY BRAZIL INDIA IORDAN KUREA .MALAYSIA MEXICO PAKISTAN. THAILAND TURKEY ZIMBADI%W IndusIrv X X* x X x X X X Listed Date . . X . X X X X _ X Eslabllir,htd Dalo X __. x _ x x x x .. Curren! Assets *.. X X X X X O X ___X__ - Nei Fixed Assets X X _X _ _ 2 __X _ X XXX Total Assets _ x X X x X _ x _ XX _ ._XXX Current liabilities .- X . x X X _X XX ..XX Lone TeFmn LiabTilites X X . X X X X _XX _ X (LTD) Total Liabilities X X. X X X X X X X X Paid in Canital X X .X X X X X X _i Raserves & Retained _ _ X X . X X X __ X X _a _ ines ____I Revaluation Fund - _ . X Total Fouilv X X X _ X X X ._ X _XX . X Sales _ x x X X x x X X X Fi'nancial Expenses - ........ L.. K. ..L... ..Q...... ..x. F BT I X x X X % X X XX EAT X X X -X X X X _ XXX Denreciation _ _ .. . - _LX Dividends X - X X X Price X K X X _ XO X X X Number of Shares X X - X X XX X % Market Value X * x . 0 X _ X 5 Years _ _ 80^90 80-90 80"9d 83 9a0Sam84-41 _ 0- &8 83 82-90 08 Number of Firms 100 ,J1 88 _38 100 100 . 100 100 69 45 48 O represents fewer data points. 38 Policy Research Working Paper Series Contact Title Author Date for paper WPS1306 Capital Flows and Long-Term Ibrahim A. Ebadawl June 1994 R. Martin Equilibrium Real Exchange Rates Raimundo Soto 39065 hI Chile WPS1307 How Taxation Affects Foreign Direct Joosung Jun June 1994 S. King-Watson Investment (Country Specific Evidence) 31047 WPSI308 Ownership and Corporate Control In Brian Pinto June 1994 M. Karrm-Cheong Poland: Why State Firms Defied the Sweder van Wijnbergen 39618 Odds WPS1309 Is Demand for Poliuting Goods Gunnar S. Eskeland June 1394 C. Jones Manageable? An Econometric Study Tarhan N. Feyzioglu 37699 of Car Ownership and Use in Mexico WPS1310 China's Economic Reforms: Pointers Justin Yifu Lin June 1994 C. Spooner for Other Economies in Transition Fang Cai 30464 Zhou U WPS1311 The Supply Response to Exchange Mustapha Rouis June 1994 J. Schwartz Rate Reform in Sub-Saharan Africa Weshah Razzak 32250 (Empirical Evidence) Cados Mollinedo WPS1312 The New Wave of Private Capital Eduardo Femandez-Arias June 1994 R. Tutt Inflows: Push or Pull? 31047 WPS1313 New Estimates of Total Factor Vikram Nehru June 1994 M. Coleridge- Productivity Growth for Developing Ashok Dhareshwar Taylor and Industrial Countries 33704 WPS1314 The Significance of the 'Europe Bartlomiej Kaminski June 1994 M. Patenia Agreementsw for Central European 37947 Industrial Exports WPS1315 Global Tradable Carbon Permins, Bjom Larsen June 1994 C. Jones Participation Incentives, and Anwar Shah 37754 Transfers WPS1316 Preserving the CFA Zone: Macroeco- Shantayanan Devarajan June 1994 C. Jones nomic Coordination After the Michael Walton 37699 Devaluation WPS1317 Estimating the Efficiency Gains Jeremy Bulow July 1994 Rose Vo of Debt Restructuring Kenneth Rogoff 33722 Ning S. Zhu WPS1318 Exchange-Rate-Based Stabilization Miguel A. Kiguel July 1994 M. Divino in Argentina and Chile: A Fresh Look Nissan Liviatan 33739 Policy Research Working Paper Series Contact TlUe Author Date for paper WPS1319 The Financial System and Public Ashi DemirgOu-Kunt July 1994 B. Moore Enterprise Reform: Concepts and Ross Levine 35261 Cases WPS1320 Capital Structures In Developing Asih Demirguq-Kunt July 1994 B. Moore Countries: Evidence from Ten Vojislav Maksimovic 35261 Countries