Comparative Analysis of Leverage in Czech and G7 Countries, Essays (university) of Management Accounting

Download Comparative Analysis of Leverage in Czech and G7 Countries and more Essays (university) Management Accounting in PDF only on Docsity! 2 Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 UDC: 658.14 Keywords: determinants of capital structure – extent of leverage – listed companies in the Czech Republic Determinants of Capital Structure Empirical Evidence from the Czech Republic Patrik BAUER* The modern theory of capital structure was established by Modigliani and Miller (1958). Thirty-seven years later, Rajan and Zingales (1995, p. 1421) stated: “Theory has clearly made some progress on the subject. We now understand the most important departures from the Modigliani and Miller assumptions that make capital structure relevant to a firm’s value. However, very little is known about the empirical relevance of the diffe- rent theories.” Similarly, Harris and Raviv (1991, p. 299) in their survey of capital struc- ture theories claimed: “The models surveyed have identified a large num- ber of potential determinants of capital structure. The empirical work so far has not, however, sorted out which of these are important in various contexts.” Thus, several conditional theories of capital structure exist (none is universal), but very little is known about their empirical relevance. More- over, the existing empirical evidence is based mainly on data from deve- loped countries (G7 countries). Findings based on data from developing countries have not appeared until recently – for example Booth et al. (2001)1 or Huang and Song (2002)2. So far, no study has been published based on data from transition countries of Central and Eastern Europe, at least to the extent of this author’s knowledge. The main goal of this paper is to fill this gap, exploring the case of the Czech Republic. The structure of this paper is as follows. In Section 1 the most prominent theoretical and empirical findings are surveyed. In Section 2 the potential determinants of capital structure are summarized and theoretical and em- pirical evidence concerning these determinants is provided. Section 3 is the empirical part of the paper. Here the data is described, measures of * Institute of Economic Studies, Charles University, Prague and IDET, Department of Eco- nomics, Ludwig-Maximilians-Universität, Munich (Patrik.Bauer@seznam.cz) This research was supported by the Grant Agency of the Czech Republic, Grant No. 402/01/0033, ‘Risk management and financial engineering: applicability of modern technology on capital market in the Czech Republic’. The author is grateful to the European Commission for sup- porting his Marie Curie Fellowship at LMU, Munich. Moreover, he would like to thank Zdenek Sid Blaha, Vít Bubák, Jana Fajtová, Irena Jindfiichovská and the participants in the research seminar at the Seminar for Comparative Economics, Munich (especially Monika Schnitzer) for their selfless help and helpful comments. 1 the cases of Brazil, Mexico, India, South Korea, Jordan, Malaysia, Pakistan, Thailand, Turkey and Zimbabwe 2 the case of China leverage are defined, the extent of leverage is characterized and the impact of potential determinants of capital structure on leverage is tested. Sec- tion 4 provides conclusions of the study. 1. Theoretical and Empirical Findings According to Myers (2001, p. 81), “there is no universal theory of the debt- -equity choice, and no reason to expect one”. However, there are several use- ful conditional theories3, each of which helps to understand the debt-to- -equity structure that firms choose. These theories can be divided into two groups – either they predict the existence of the optimal debt-equity ratio for each firm (so-called static trade-off models) or they declare that there is no well-defined target capital structure (pecking-order hypothesis). Static trade-off models understand the optimal capital structure as an op- timal solution of a trade-off, for example the trade-off between a tax shield and the costs of financial distress in the case of trade-off theory. According to this theory the optimal capital structure is achieved when the marginal present value of the tax shield on additional debt is equal to the marginal present value of the costs of financial distress on additional debt. The trade- -off between the benefits of signaling and the costs of financial distress in the case of signaling theory implies that a company chooses debt ratio as a signal about its type. Therefore in the case of a good company the debt must be large enough to act as an incentive compatible signal, i.e., it does not pay off for a bad company to mimic it. In the case of agency theory the trade-off between agency costs4 stipulates that the optimal capital struc- ture is achieved when agency costs are minimized. Finally, the trade-off be- tween costs of financial distress and increase of efficiency in the case of free cash-flow theory, which is designed mainly for firms with extra-high free cash-flows, suggests that the high debt ratio disciplines managers to pay out cash instead of investing it below the cost of capital or wasting it on or- ganisational inefficiencies. On the other hand, the pecking-order theory sug- gests that there is no optimal capital structure. Firms are supposed to pre- fer internal financing (retained earnings) to external funds. When internal cash-flow is not sufficient to finance capital expenditures, firms will bor- row, rather than issue equity. Therefore there is no well-defined optimal leverage, because there are two kinds of equity, internal and external, one at the top of the pecking order and one at the bottom.5 Existing empirical evidence is based mainly on data from developed coun- tries. For example Bradley et al. (1984), Kim and Sorensen (1986), Friend 3Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 3 The most prominent theories are the trade-off theory, the signaling theory (first mentioned by (Ross, 1977)), the agency theory (Jensen – Meckling, 1976), (Myers, 1977), the free cash-flow theory (Jensen, 1986), and the pecking-order theory (Myers – Majluf, 1984), (Myers, 1984). For more details about conditional capital structure theories see (Bauer – Bubák, 2003). 4 Agency costs arise from two agency relations – the relation between owners and debt hold- ers and the relation between owners and managers (non-owners), i.e., principal-agent rela- tion. 5 For a comprehensive survey of capital structure theories see (Harris – Raviv, 1991), for theo- ries based on asymmetric information see (Klein et al., 2002). et al. (2001, p. 101) state: “The more tangible the firm’s assets, the greater its ability to issue secured debt and the less information revealed about fu- ture profits.” Thus a positive relation between tangibility and leverage is predicted. Several empirical studies confirm this suggestion, such as (Rajan – Zin- gales, 1995), (Friend – Lang, 1988) and (Titman – Wessels, 1988) find. On the other hand, for example Booth et al. (2001) and Huang and Song (2002) experience a negative relation between tangibility and leverage. In this study, tangibility is defined as tangible assets divided by total as- sets. 2.4 Growth Opportunities According to Myers (1977), firms with high future growth opportunities should use more equity financing, because a higher leveraged company is more likely to pass up profitable investment opportunities. As Huang and Song (2002, p. 9) claim: “Such an investment effectively transfers wealth from stockholders to debtholders.” Therefore a negative relation between growth opportunities and leverage is predicted. As market-to-book ratio is used in order to proxy for growth opportunities, there is one more reason to expect a negative relation – as Rajan and Zingales (1995, p. 1455) point out: “The theory predicts that firms with high market-to-book ratios have higher costs of financial distress, which is why we expect a negative corre- lation.” Some empirical studies confirm the theoretical prediction, such as (Ra- jan – Zingales, 1995), (Kim – Sorensen, 1986) or (Titman – Wessels, 1988) report. However, for example, Kester (1986) and Huang and Song (2002) demonstrate a positive relation between growth opportunities and lever- age. In this study, the P/B ratio (market-to-book ratio) is used as a proxy for growth opportunities. 2.5 Tax According to the trade-off theory, a company with a higher tax rate should use more debt and therefore should have higher leverage, because it has more income to shield from taxes. However, for example Fama and French (1998) declare that debt has no net tax benefits. As MacKie-Mason (1990, p. 1471) claims: “Nearly everyone believes taxes must be important to fi- nancing decision, but little support has been found in empirical analysis.” As he also points out (MacKie-Mason, 1990, p. 1471): “This paper provides clear evidence of substantial tax effects on the choice between issuing debt or equity; most studies fail to find significant effects. [...] Other papers miss the fact that most tax shields have a negligible effect on the marginal tax rate for most firms. New predictions are strongly supported by an empi- rical analysis; the method is to study incremental financing decisions us- ing discrete choice analysis. Previous researchers examined debt-equity ratios, but tests based on incremental decisions should have greater power.” As he adds, debt-equity ratios “are the cumulative result of years of sepa- 6 Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 rate decisions. Tests based on a single aggregate of different decisions are likely to have low power for effects at the margin.” (MacKie-Mason, 1990, p. 1472). However, as data to perform similar analysis as (MacKie-Mason, 1990) is not available in the Czech Republic, the average tax rate defined as the dif- ference between earnings before taxes and earnings after taxes, scaled by earnings before taxes, is used as a proxy variable to analyse the tax effects on leverage in this study. 2.6 Non-debt Tax Shields Other items apart from interest expenses, which contribute to a de- crease in tax payments, are labelled as non-debt tax shields (for exam- ple the tax deduction for depreciation). According to Angelo – Masulis (1980, p. 21): “Ceteris paribus, decreases in allowable investment-re- lated tax shields (e.g., depreciation deductions or investment tax cre- dits) due to changes in the corporate tax code or due to changes in in- flation which reduce the real value of tax shields will increase the amount of debt that firms employ. In cross-sectional analysis, firms with lower investment related tax shields (hol-ding before-tax earnings constant) will employ greater debt in their capital structures.” So they argue that non-debt tax shields are substitutes for a debt-related tax shield and therefore the relation between non-debt tax shields and lever- age should be negative. Some empirical studies confirm the theoretical prediction, for example Kim and Sorensen (1986, p. 140) declare: “DEPR9 has a significantly nega- tive coefficient. [...] This is consistent with the notion that depreciation is an effective tax shield, and thus offsets the tax shield benefits of leverage.” A negative relation between non-debt tax shields and leverage is also found by (Huang – Song, 2002) and (Titman – Wessels, 1988). However, for exam- ple Bradley et al. (1984) and Chaplinsky and Niehaus (1993) observe a po- sitive relationship between non-debt tax shields and leverage. Depreciation divided by total assets is used in order to proxy for non-debt tax shields in this study. 2.7 Volatility Volatility may be understood as a proxy for risk of a firm (probability of bankruptcy). Therefore it is assumed that volatility is negatively related to leverage. However, as Huang and Song (2002, p. 9) state based on findings of Hsia (1981): “As the variance of the value of the firm’s assets increases, the systematic risk of equity decreases. So the business risk is expected to be positively related to leverage.” The positive relation between volatility and leverage is confirmed by (Kim 7Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 9 Kim and Sorensen (1986, p. 138) define DEPR as the average rate of depreciation, i.e., depre- ciation charges divided by fixed assets. – Sorensen, 1986) and (Huang – Song, 2002). Conversely, a negative rela- tion is found by (Bradley et al., 1984) and (Titman – Wessels, 1988). In this study, standard deviation of return on assets is used as a proxy for volatility. 2.8 Industry Classification Some empirical studies identify a statistically significant relationship be- tween industry classification and leverage, such as (Bradley et al., 1984), (Long – Malitz, 1985), and (Kester, 1986). As Harris and Raviv (1991, p. 333) claim, based on a survey of empirical studies: “Drugs, Instruments, Elec- tronics, and Food have consistently low leverage while Paper, Textile Mill Products, Steel, Airlines, and Cement have consistently large leverage.” To estimate the effect of industry classification on leverage, firms in our sample are divided into groups according to the Industrial Classification of Economic Activities of the Czech Statistical Office. The following classifi- cation is used in order to create reasonably large groups of firms: C – Mi- ning of Raw Materials, D – Manufacturing, and E – Production and Dis- tribution of Electricity, Gas, and Water. Firms not belonging to any of these groups make up the reference group. Titman (1984) and Titman and Wes- sels (1988) point out that firms manufacturing machines and equipment should be financed with relatively less debt. Because group D is sufficiently large, it is possible to drop the firms that belong to sub-industry 29 (Manu- facturing and Repair of Machines and Equipment) and create from these firms the group D1. Therefore four dummy variables are used in the empirical analysis to es- timate the effect of industry classification on leverage – C, D, D1, and E. Table 1 provides a brief summary of theoretical and empirical findings (except for industry dummies). 3. Empirical Analysis 3.1 Data Description Data used in the analysis were collected from financial reports of com- panies as available on the Prague Stock Exchange website and in the Se- curities Centre of the Czech Republic data base; prices of ordinary shares at year-end were obtained from “Burzovní noviny”, the official stock mar- ket supplement to the “Hospodáfiské noviny” daily. The data are based on financial reports according to Czech Accounting Standards (the only data available for all companies). The sample comprises a total of 74 companies listed on the Prague Stock Exchange within the period from 2000 to 2001 (i.e., all non-financial com- panies traded on the Prague Stock Exchange in November 2002). After a data set investigation it was decided to exclude 2 companies – one because of negative book value and one because of unusual changes in ba- lance sheet items between the years 2000 and 2001. Therefore the final sample comprises 72 observations for each year. 8 Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 G7 countries. The TD value for developing countries is not available except for China, which exhibits a lower TD value than G7 countries (and com- pared to the Czech Republic). Different results are obtained when leverage is measured in market value. Because of a very low P/B ratio in the Czech Republic (for more details, see below), the value of Market Total Liabilities Ratio (MTL) in the case of the Czech Republic is higher than in G7 count- ries (only Italy exhibits a higher MTL value, France shows a similar value). China presents a very special case; the reason is given below where P/B ra- tios are discussed. Measured by Market Total Debt Ratio (MTD), the value of leverage is much higher in the Czech Republic than in G7 countries or China (which is again a very special case). Therefore the Czech Republic shows relatively low leverage if measured in book value, but relatively high if measured in market value. 3.3 Determinants of Capital Structure As stated in Chapter 2, eight possible determinants of leverage are ana- lysed in this study. They are summarized in Table 5, descriptive statistics of both dependent and explanatory variables and the size of industry groups are reported in Table 6, and an international comparison of descriptive statistics of explanatory variables (except for industry dummies) is shown in Table 7. 11Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 Country Obs. Period TL TD MTL MTD Czech Republic 72 2000 0.44 (0.42) 0.41 (0.39) 0.65 (0.62) 0.60 (0.59) Czech Republic 72 2001 0.40 (0.41) 0.36 (0.37) 0.66 (0.62) 0.62 (0.58) USA 2580 1991 0.58 (0.66) 0.37 (0.37) 0.44 (0.44) 0.28 (0.32) Japan 514 1991 0.69 (0.67) 0.53 (0.52) 0.45 (0.45) 0.29 (0.31) Germany 191 1991 0.73 (0.72) 0.38 (0.39) 0.60 (0.56) 0.23 (0.28) France 225 1991 0.71 (0.69) 0.48 (0.46) 0.64 (0.61) 0.41 (0.41) Italy 118 1991 0.70 (0.67) 0.47 (0.46) 0.70 (0.67) 0.46 (0.47) UK 608 1991 0.54 (0.57) 0.28 (0.29) 0.40 (0.42) 0.19 (0.23) Canada 318 1991 0.56 (0.61) 0.39 (0.39) 0.49 (0.47) 0.35 (0.36) China 954 2000 0.45 (0.46) 0.28 (0.31) 0.12 (0.14) 0.06 (0.08) Brazil 49 85–91 0.30 N/A N/A N/A Mexico 99 84–90 0.35 N/A N/A N/A India 99 80–90 0.67 N/A N/A N/A South Korea 93 80–90 0.73 N/A N/A N/A Jordan 38 83–90 0.47 N/A N/A N/A Malaysia 96 83–90 0.42 N/A N/A N/A Pakistan 96 80–87 0.66 N/A N/A N/A Thailand 64 83–90 0.49 N/A N/A N/A Turkey 45 83–90 0.59 N/A N/A N/A Zimbabwe 48 80–88 0.42 N/A N/A N/A TABLE 4 Extent of Leverage in Selected Countries, Median (Mean) Source: Czech Republic: own calculations; G7 countries: (Rajan – Zingales, 1995); China: (Huang – Song, 2002); other countries: (Booth et al., 2001) The explanation of descriptive statistics is best in the context of an in- ternational comparison as shown in Table 7. In general, the firm SIZE is higher in G7 countries than in developing countries including the Czech Republic. Among G7 countries, values of SIZE are very similar. The value of SIZE in the Czech Republic is higher than in most other developing countries. In the Czech Republic, the value of ROA is slightly lower than in G7 countries and much lower than in developing countries. The value of ROA is 1.5 points higher in developing countries than in G7 countries. The highest profitability is exhibited by firms in Thai- land and Zimbabwe; in G7 countries firms in the UK show the highest profi- tability. Tangibility in the Czech Republic is higher than in G7 countries except Canada, where the value is similar. Only in two developing count- ries, Brazil and Malaysia, is the tangibility higher than in the Czech Re- public. In general, tangibility is higher in developing countries than in de- veloped countries. 12 Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 SIZE Size ln Sales ROA Profitability EBIT/TA TANG Tangibility Tangible Assets/TA PB Growth opportunities P/B ratio TAX Average tax rate (EBT – E)/EBT NDTS Non-debt tax shields Depreciation/TA VOLTY Volatility SD (ROA) IND_C Industry dummy = 1 if the firm belongs to Mining of Raw Materials; = 0 otherwise IND_D Industry dummy = 1 if the firm belongs to Manufacturing, except for Manufacturing and Repair of Machines and Equipment; = 0 otherwise IND_D1 Industry dummy = 1 if the firm belongs to Manufacturing and Repair of Machines and Equipment; = 0 otherwise IND_E Industry dummy = 1 if the firm belongs to Production and Distribution of Electricity, Gas, and Water; = 0 otherwise TABLE 5 Determinants of Capital Structure Note: TAX is defined as TAX = 0 in cases where the value of TAX is outside the interval <0;1> (for example when both EBT and E are negative and EBT>E). This happened in 10 cases in 2000, and in 3 cases in 2001. 2000 Obs. Mean Median Min. Max. SD 2001 Obs. Mean Median Min. Max. SD TL 72 0.42 0.44 0.07 0.96 0.19 TL 72 0.41 0.40 0.01 0.95 0.20 TD 72 0.39 0.41 0.00 0.95 0.20 TD 72 0.37 0.36 0.00 0.94 0.21 MTL 72 0.62 0.65 0.10 0.98 0.23 MTL 72 0.62 0.66 0.02 0.99 0.23 MTD 72 0.59 0.60 0.00 0.97 0.24 MTD 72 0.58 0.62 0.02 0.99 0.25 SIZE 72 21.47 21.84 11.78 24.71 2.17 SIZE 72 21.52 21.89 11.87 24.68 2.16 ROA 72 0.06 0.05 –0.21 0.71 0.12 ROA 72 0.05 0.05 -0.27 0.37 0.09 TANG 72 0.53 0.59 0.00 0.95 0.25 TANG 72 0.52 0.55 0.00 0.93 0.26 PB 72 0.53 0.39 0.02 2.11 0.43 PB 72 0.47 0.37 0.03 2.25 0.38 TAX 72 0.21 0.24 0.00 0.66 0.18 TAX 72 0.22 0.26 0.00 0.74 0.18 NDTS 72 0.05 0.05 0.00 0.10 0.03 NDTS 72 0.05 0.05 0.00 0.12 0.03 VOLTY 72 0.03 0.01 0.00 0.31 0.04 VOLTY 72 0.03 0.01 0.00 0.31 0.04 Obs (IND_C) = 8 Obs (IND_D) = 17 Obs (IND_D1) = 8 Obs (IND_E) = 25 TABLE 6 Descriptive Statistics The most interesting comparison concerns the value of PB. In the Czech Republic, the P/B ratio is the lowest of all the countries reported in Table 7. Three other developing countries, South Korea, Pakistan and Zimbabwe, also exhibit a P/B ratio less than one. There is the opposite situation in all G7 countries, where the lowest P/B ratio is exhibited by Italy (exactly one). In all other countries the P/B value is greater than one. China is a very spe- cial case, the average value of P/B ratio is over five; therefore leverage in market value is much lower than leverage in book value. The value of TAX is not available for G7 countries except for the USA, where the value is higher than that of the Czech Republic. In developing countries, in four cases, the value of TAX is lower. It is the same in the case of India and in six other cases it is higher than in the Czech Republic. The value of NDTS is available only for two other countries. A sensible com- parison is, therefore, not possible. However, the value of NDTS is highest in USA and lowest in China. The value of VOLTY is lower in the Czech Re- public than in G7 countries, except for Japan; the highest value is exhibi- ted by the USA. Concerning developing countries, the value of VOLTY in the Czech Republic is less than or equal to any of them. The value of VOLTY 13Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 SIZE (USD) ROA TANG PB TAX NDTS VOLTY CR 2000 17.98 (2.17) 0.06 (0.12) 0.53 (0.25) 0.53 (0.43) 0.21 (0.18) 0.05 (0.03) 0.03 (0.04) CR 2001 18.02 (2.16) 0.05 (0.09) 0.52 (0.26) 0.47 (0.38) 0.22 (0.18) 0.05 (0.03) 0.03 (0.04) USA 21.61 0.07 (0.08) 0.36 1.65 0.30 (0.18) 0.10 (0.04) 0.07 (0.12) Japan 21.83 0.05 (0.03) 0.29 1.67 N/A N/A 0.02 (0.04) UK 20.59 0.09 (0.08) 0.41 1.35 N/A N/A 0.06 (0.09) Germany 21.65 0.06 (0.04) 0.33 1.57 N/A N/A 0.04 (0.06) France 21.69 0.07 (0.05) 0.24 1.26 N/A N/A 0.04 (0.05) Italy N/A N/A 0.32 1.00 N/A N/A N/A Canada N/A N/A 0.52 1.36 N/A N/A N/A China 19.7 (1.0) 0.08 (0.04) 0.34 (0.16) 3.19 (1.24) 0.16 (0.06) 0.02 (0.02) 0.04 (0.04) Brazil 13.1 (1.0) 0.07 (0.12) 0.68 (0.19) N/A 0.14 (0.17) N/A 0.09 (0.05) Mexico 11.2 (1.4) 0.08 (0.08) 0.33 (0.30) N/A 0.26 (0.57) N/A 0.06 (0.03) India 18.4 (1.0) 0.07 (0.07) 0.41 (0.18) 1.4 (1.1) 0.22 (0.21) N/A 0.05 (0.03) South Korea 18.9 (0.9) 0.04 (0.04) 0.49 (0.15) 0.7 (0.7) 0.30 (0.20) N/A 0.03 (0.02) Jordan 9.8 (0.3) 0.07 (0.11) 0.47 (0.22) 1.4 (0.7) 0.16 (0.18) N/A 0.08 (0.04) Malaysia 17.4 (1.6) 0.07 (0.07) 0.58 (0.22) 2.3 (1.8) 0.32 (0.44) N/A 0.05 (0.03) Pakistan 17.1 (1.1) 0.09 (0.10) 0.38 (0.20) 0.9 (0.7) 0.12 (0.20) N/A 0.06 (0.04) Thailand 16.7 (1.3) 0.13 (0.07) 0.36 (0.17) 3.2 (2.1) 0.29 (0.09) N/A 0.03 (0.03) Turkey 17.2 (1.7) 0.10 (0.09) 0.41 (0.19) 1.9 (1.3) 0.30 (0.19) N/A 0.06 (0.03) Zimbabwe 16.7 (1.6) 0.12 (0.09) 0.44 (0.13) 0.6 (0.6) 0.29 (0.21) N/A 0.06 (0.06) TABLE 7 International Comparison of Descriptive Statistics, Mean (standard deviation) Source: Czech Republic: own calculations; G7 countries: see notes below; China: (Huang – Song, 2002); other count- ries: (Booth et al., 2001) Notes: Number of observations and the year under analysis are the same as reported in Table 4. PB in the case of China means Tobin’s Q (defined as market-to-book ratio of total assets); the corresponding value of P/B ra- tio is 5.24 ((calculated from (Huang – Song, 2002)). TANG in the case of (Booth et al., 2001) is defined as to- tal assets less current assets divided by total assets. Values for G7 countries were obtained as follows – TANG and PB were calculated from (Rajan – Zingales, 1995); VOLTY, ROA and SIZE from (Wald, 1999), where the same time period as in (Rajan – Zingales, 1995) is used, SIZE is defined as ln (total assets). TAX and NDTS for USA are from (Kim – Sorensen, 1986), however, the values are from the period 1975–1980. nificant in two unrestricted model and in four restricted models. The sign is always negative. Theoretically, the expected relationship between growth opportunities and leverage is negative. The results of this study confirm this expectation, as do the empirical studies for developed countries. A rather positive rela- tion is detected in developing countries. However, PB is highly statistically significant only in cases when leverage is expressed in market value. The theory expects a positive impact of taxes on leverage. However, em- pirical findings are not clear. This is equally true for the results of this stu- dy. TAX is statistically significant in two unrestricted and two restricted models, in which it shows a positive sign. Nevertheless, the relationship between TAX and leverage exhibits rather a negative sign in models for the year 2001. However, it is statistically insignificant. For non-debt tax shields, the results confirm theoretical prediction, i.e., 16 Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 2000 2001 TL TD MTL MTD TL TD MTL MTD N 70 71 68 71 72 72 69 69 Intercept -0.635*** -0.617*** -0.282** -0.331* -0.798*** -0. 761*** -0.380** -0.378* t (N–k) (-3.88) (-3.46) (-2.03) (-1.84) (-4.27) (-3.89) (-2.50) (- 1.99) SIZE 0.054*** 0.050*** 0.058*** 0.056*** 0.066*** 0.061*** 0.059*** 0.057*** (6.58) (5.51) (8.05) (6.02) (6.93) (6.24) (7.75) (6.15) ROA -0.409*** -0.367** -0.380* -0.519** -0.691** (-2.89) (-2.42) (-1.71) (-2.34) (-2.57) TANG -0.228*** -0.165*** -0.355*** -0.330*** (-3.32) (-2.67) (-4.22) (-3.87) PB -0.467*** -0.397*** -0.098* -0.341*** -0.298*** (-12.69) (-9.06) (-1.98) (-6.79) (-4.89) TAX 0.265*** 0.277** (2.79) (2.57) NDTS -2.164*** -1.573* -1.558** -1.978** (-2.82) (-1.94) (-2.64) (-2.77) VOLTY IND_C -0.145*** -0.132** -0.113* -0.109* -0.133** (-2.89) (-2.39) (-1.95) (-1.81) (-2.10) IND_D 0.105** 0.064* 0.086* (2.41) (1.77) (1.97) IND_D1 0.124** (2.16) IND_E 0.098** 0.093** (2.22) (2.03) Adj. R 2 0.50 0.46 0.74 0.62 0.44 0.41 0.68 0.59 TABLE 10 OLS Estimation Results of Restricted Models Notes: *** = 1%,** = 5%,* = 10% level of significance. Figures under the values estimated represent t-statistics (in parentheses). Results were obtained as follows. First, outliers were identified and removed from the analysis (outliers were defined as the cases in which the standard residual value, i.e., the difference between the obser- ved and predicted value divided by the square root of the residual mean square, was greater than
2 times the standard deviation). Consequently the least significant regressor was removed after each run until all regressors were statistically significant at the 10% level. a negative relation to leverage. NDTS is not statistically significant in any unrestricted model. However, it is significant in four restricted models. The sign is always negative. Theoretical prediction about the relation of volatility and leverage is not clear. This study does not provide us with a clear empirical result, because VOLTY is not statistically significant in any model. As far as industry classification is concerned, the results show a statisti- cally insignificant relationship between industry dummies and leverage in unrestricted models. The restricted models reveal that firms belonging to industry C (Mining of Raw Materials) demonstrate lower leverage, whereas firms belonging to industries D (Manufacturing except for Manufacturing and Repair of Machines and Equipment), D1 (Manufacturing and Repair of Machines and Equipment), and E (Production and Distribution of Electri- city, Gas, and Water) exhibit larger leverage than firms belonging to the re- ference group. According to the values of adjusted coefficient of determination (Adj. R2), the explanatory power of models is higher when the leverage is expressed in market value than in book value. The explanatory power of models pre- sented in this study is, in general, relatively high compared to studies of similar character. Based on the fact that two measures of leverage are used in this study, each of them in book value and in market value, that unrestricted and re- stricted models are presented, and that two cross-sectional data samples, i.e., data for the years 2000 and 2001, are used for analysis, the results prove to be quite robust. For greater clarity, the results are summarized in Table 11. 4. Conclusions In this paper, the determinants of capital structure of listed companies in the Czech Republic are analysed. In general, Czech listed firms exhibit lower leverage than firms in G7 count- 17Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 Regressor Sign Significant regressors SIZE + ROA – TANG – PB – Less significant TAX + NDTS – IND_C – IND_D + IND_D1 + IND_E + Insignificant VOLTY TABLE 11 General Results Note: PB is statistically significant only when leverage is expressed in market value. ries and firms in the majority of developing countries11, when measured by Book Total Liabilities Ratio. When evaluated by Book Total Debt Ratio, Czech companies show similar leverage as companies in G7 countries. Dif- ferent results are obtained when leverage is expressed in market value. Be- cause of very low P/B ratio, leverage in the Czech Republic is higher than in G7 countries. Thus, Czech firms show relatively low leverage measured in book value, but high leverage assessed in market value. Based on data availability, eight potential determinants of capital struc- ture were analysed in this paper – size, profitability, tangibility, growth op- portunities, tax, non-debt tax shields, volatility, and industry classification. Several interesting findings can be derived from an international com- parison of values of proxy variables for the above-mentioned determinants of leverage. Firms are generally bigger in G7 countries than in developing countries including the Czech Republic. Concerning Czech firms, they are generally bigger than firms in most developing countries. Firms in the Czech Republic are slightly less profitable than firms in G7 countries and much less profitable than firms in developing countries. Tangibility in the Czech Republic is higher than in G7 countries and in most developing countries. The P/B ratio in the Czech Republic is the lowest among all countries as reported in this paper. Three reported developing countries (except for the Czech Republic) exhibit a P/B ratio less than one. The opposite situa- tion exists in all G7 countries, where the P/B ratio is greater than or equal to one. In the USA, the average tax rate is higher than in the Czech Re- public. The average tax rate is lower in four developing countries reported in this study than in the Czech Republic, in one case it is the same and in six cases it is higher. The value of non-debt tax shields is available only for two other countries, therefore no sensible comparison is possible. The profit volatility is on average lower in the Czech Republic than in G7 countries. Concerning the developing countries, profit volatility in the Czech Repub- lic is less than or equal to any of them. According to the results of empirical analysis, leverage of Czech listed firms is positively correlated with size, this result supports the view of size as an inverse proxy for the probability of bankruptcy. Leverage is negatively correlated with profitability. This finding is consistent with the pecking or- der hypothesis rather than with static trade-off models. A negative relation between tangibility and leverage is in contradiction with theoretical pre- diction. The reason for this result would need some theoretical support, which is not provided by this study. The relationship between leverage and P/B ratio (proxy for growth opportunities) is negative, given that the leve- rage is measured in market value. This result confirms that firms with higher future growth opportunities should use more equity financing. It can be stated, on the lower level of statistical significance, that leve- rage is positively correlated with tax, and dummy variables for Manufac- turing except for Manufacturing and Repair of Machines and Equipment, Manufacturing and Repair of Machines and Equipment, and Production and Distribution of Electricity, Gas, and Water, and it is negatively corre- 18 Finance a úvûr – Czech Journal of Economics and Finance, 54, 2004, ã. 1-2 11 “Developing countries” comprise eleven developing countries reported in this study.