Firm, country and macroeconomic determinants of capital structure: Evidence from transition economies

Journal of Comparative Economics 41 (2013) 294–308

Contents lists available at SciVerse ScienceDirect

Journal of Comparative Economics journal homepage: www.elsevier.com/locate/jce

Firm, country and macroeconomic determinants of capital structure: Evidence from transition economies Karin Jõeveer Management School, Queen’s University Belfast, BT9 5EE, United Kingdom

a r t i c l e

i n f o

Article history: Received 7 April 2011 Revised 1 May 2012 Available online 18 May 2012 JEL classification: G32 Keywords: Capital structure Eastern Europe

a b s t r a c t Jõeveer, Karin—Firm, country and macroeconomic determinants of capital structure: Evidence from transition economies This study explores the significance of firm-specific, institutional, and macroeconomic factors in explaining variation in leverage using a sample of firms from nine Eastern European countries. Country-specific factors are the main determinants of variation in leverage for small unlisted companies, while firm-specific factors explain most of the variation in leverage for listed and large unlisted companies. Around half of the variation in leverage related to country factors is explained by known macroeconomic and institutional factors, while the remainder is explained by unmeasurable institutional differences. Journal of Comparative Economics 41 (1) (2013) 294–308. Management School, Queen’s University Belfast, BT9 5EE, United Kingdom. Ó 2012 Association for Comparative Economic Studies Published by Elsevier Inc. All rights reserved.

1. Introduction Firm capital structure is irrelevant in efficient financial markets as shown by Modigliani and Miller (1958).1 Subsequent theoretical work has taken into account the imperfections of financial markets and has shown that firm capital structure emerges from firm-specific and macroeconomic factors.2 In their pioneering exploration of capital structure around the world Rajan and Zingales (1995) observe similar levels of leverage for firms from G-7 countries. They conclude that it is difficult to explain the leverage differences by country institutional differences and stress that ‘‘a better understanding of the influence of institutions can provide us enough intercountry variation so as to enable us to identify the fundamental determinants of capital structure’’. In this paper we pursue this agenda by first documenting that country of incorporation in itself accounts for much of the variation in leverage unexplained by firm-specific or macroeconomic factors. We use a large sample of firms from several Eastern European countries covering the time period encompassing the transition from socialist planning to a market-based capitalistic system. We then identify institutional variables which in turn account for some of this country effect. Finally, by examining capital structure early and late in the transition process we see to what extent institutional factors account for changes in capital structure. The importance of the country of incorporation for firm leverage has been analysed in several earlier cross-country studies. Booth et al. (2001) show on a sample of firms from ten developing countries that country fixed effects explain a large

1 2

E-mail address: [email protected] Rubinstein (2003) notes that Williams (1938) already expressed the same idea. For a recent empirical capital structure study and for references to the literature see Frank and Goyal (2009).

0147-5967/$ - see front matter Ó 2012 Association for Comparative Economic Studies Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jce.2012.05.001

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

295

share of leverage variation, but they do not decompose the country effects to show what country characteristics matter. On a sample of firms from developing Asian and South American countries, Schmukler and Vesperoni (2001) explore the relation between leverage and financial liberalization. Using data on Western European firms Giannetti (2003) shows that financial development and creditor protection are significant determinants of leverage. Jõeveer (2005), also using Western European firm data, shows that half of the country explanatory power is determined by measurable country macroeconomic and institutional factors while another half is explained by an unmeasurable institutional differences. De Jong et al. (2008) argue by using firms from both developing and developed countries that country factors matter to the firm capital structure and the effect can be either direct or indirect through the firm-specific determinants. Based on small and medium-sized enterprises (SME)3 from four Western European countries Psillaki and Daskalakis (2009) acknowledge the different relations between capital structure determinants and leverage across countries but they conclude that firm factors rather than country factors explain the differences in capital structure choices among the firms. In addition Bancel and Mittoo (2004) surveyed the managers from 16 European countries firms and found support for the importance of institutional factors on capital structure choices. In this paper we employ firm-level data from nine Eastern European countries over the period 1995–2002. With the collapse of the Soviet block in the late 1980’s the firms in those countries had to readjust their working principles for being competitive in the open market. Also it brought along a wave of newborn enterprises. Those countries were going through severe economic reforms during 1990’s producing a change in the institutional settings as well as a change in macroeconomic indicators. Berglof and Bolton (2002) record that the methods and speed with which the missing institutions were introduced differed across countries, providing additional time-variation in country factors. All these major changes were expected to have an impact on the capital structure of the firm and make firms from Eastern Europe particularly interesting to study for understanding the institutional underpinnings of firm capital structure. The importance of studying the capital structure of firms in transition economies was first pointed out by Cornelli et al. (1998). In addition there are several cross-country studies based on Eastern European firms (Nivorozhkin, 2005; De Haas and Peeters, 2006; Decoure, 2007). The current study complements these existing studies in using a larger, more complete data set and by investigating in detail a variety of institutional characteristics that may account for country of incorporation effects. We start with an analysis of variance which is helpful in documenting the importance of country effects as opposed to industry and time effects in understanding capital structure differences in transition countries. Next we incorporate firmspecific factors into the analysis and observe that country effects continue to be highly significant. We then ask whether there are observable proxies that account for these country-based institutional differences and find that some but not all of the country effect can be explained in this way. Finally, we study the evolution of capital structure during the transition process and ask whether institutional factors account for observed changes in leverage. The paper is organized as follows: In the next section we provide an overview of the related research. In Section 3 we introduce the data and the estimation strategy. Section 4 contains the results, followed by a concluding section.

2. What do we know about capital structure in transition economies? The two most influential theories of capital structure—trade-off theory (TOT) and pecking order theory (POT)—offer several predictions regarding to firm-specific and country-specific factors affecting firm leverage. Trade-off theory argues that firms balance the tax benefits of debt with potential bankruptcy costs to achieve an optimal leverage level (see Miller (1977) for a discussion). Given this the TOT predicts that larger firms, firms with more tangible assets and with higher profitability could enjoy larger tax benefits of debt and hence should have higher leverage. In addition, local tax levels as well as bankruptcy codes matter in principle. Higher tax rates imply greater interest tax shields benefits and therefore induce higher leverage. The expected inflation is predicted to be positively related to leverage due to higher real value of tax deductions on debt. The interest rates, as a proxy for the cost of debt, should be negatively related to leverage. All this predictions should be applicable also in the case of Eastern European firms. The pecking order theory of capital structure argues that, firms prefer internal funds to outside sources since the latter are mispriced due to the asymmetric information between owners and investors (see Myers, 1984). Based on this, the POT predicts that more profitable firms can rely on internal funds and hence have lower leverage. POT stresses the importance of the transparency of the firm activities. Given the newness of stock markets and general opacity of business connections in transition countries, problems of asymmetric information are expected to be especially large, meaning that firms are less likely to turn to outside sources of finance even if the investment opportunities exceed the internal funds. Many institutional changes during the transition process were designed to enhance transparency in Eastern Europe. The different economic policies employed by those countries might explain the large differences in the levels and changes in the capital structure of firms experienced across those countries (see Table 1 Panels A and B). In additional to TOT and POT theoretical literature of capital structure has proposed the possibility that some capital structure decisions may be motivated by firms’ attempts to ‘‘time the market’’, i.e., to issue debt or equity when times are particularly propitious (see Baker and Wurgler, 2002). Since firms are relatively young and small in Eastern Europe the use of external financing directly from the market is very limited and hence we would predict that business cycle should 3

European Commission defines SME as a firm with less than 250 employees and 27 million Euro in total assets.

296

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

Table 1 Summary statistics. Bulgaria

Czech Republic

Estonia

Hungary

Latvia

Lithuania

Poland

Romania

Slovakia

Panel A – across countries Broad leverage Mean 0.54 Median 0.51 St. dev. 0.35

0.59 0.59 0.31

0.61 0.63 0.28

0.62 0.62 0.25

0.64 0.68 0.29

0.56 0.57 0.23

0.63 0.64 0.28

0.68 0.71 0.32

0.57 0.57 0.32

Narrow leverage Mean Median St. dev.

0.12 0.00 0.25

0.29 0.18 0.33

0.31 0.22 0.32

0.03 0.00 0.13

0.36 0.28 0.35

0.35 0.31 0.25

0.36 0.31 0.31

0.18 0.00 0.29

0.28 0.15 0.33

Tangibility Mean Median St. dev.

0.41 0.40 0.27

0.37 0.36 0.27

0.38 0.36 0.27

0.38 0.36 0.23

0.35 0.31 0.25

0.42 0.41 0.24

0.45 0.45 0.25

0.37 0.35 0.25

0.42 0.44 0.27

Profitability Mean Median St. dev.

0.06 0.04 0.22

0.07 0.05 0.14

0.09 0.07 0.20

0.10 0.08 0.14

0.10 0.07 0.19

0.07 0.05 0.12

0.07 0.06 0.14

0.18 0.12 0.27

0.05 0.03 0.12

Total assets Mean Median St. dev.

1701 208 19,236

GDP growth Mean Median St. dev.

12,939 2688 93,673

1759 360 12,571

8355 1101 68,569

3788 762 21,688

11,000 1737 54,299

21,201 4328 121,003

2556 180 363,324

8030 2251 65,947

1.05 3.36 5.47

2.18 2.06 2.45

5.77 5.80 3.09

3.62 4.00 1.47

5.06 5.59 3.05

4.70 5.53 2.99

4.39 4.45 2.26

1.16 2.30 4.86

4.09 4.40 1.65

160.85 14.49 364.96

6.21 6.63 3.47

10.91 6.97 9.71

14.84 12.11 7.95

8.14 3.66 8.61

10.20 3.19 14.42

12.44 10.93 8.43

54.18 42.25 42.07

7.72 7.01 2.88

47.04 57.35 29.71

49.73 59.92 21.17

41.44 46.43 16.54

67.56 69.25 9.23

45.07 45.18 10.21

34.51 39.02 16.31

47.75 44.00 19.35

49.78 50.16 20.73

49.00 42.37 14.87

0.69 0.64 0.16

0.77 0.79 0.08

0.98 0.98 0.01

0.65 0.64 0.05

0.54 0.55 0.02

0.88 0.88 0.04

0.56 0.55 0.12

0.82 0.75 0.15

0.70 0.69 0.05

12.75 12.00 1.04

8.13 7.00 1.55

15.75 15.00 1.04

7.50 6.00 2.07

15.00 15.00 0.00

18.75 18.00 1.04

16.50 18.00 2.07

12.13 11.00 1.55

8.00 8.00 0.00

Minority shareholder rights Mean 8.50 Median 7.00 St. dev. 2.07

10.75 10.00 1.04

7.75 7.00 1.04

10.25 8.00 3.11

9.00 9.00 0.00

11.38 11.00 0.52

16.25 17.00 1.04

10.13 9.00 1.55

12.00 12.00 0.00

Corporate income tax Mean 23.25 Median 24.00 St. dev. 3.69

31.00 31.00 6.23

28.50 29.00 1.77

18.50 18.00 0.93

16.88 18.00 1.55

16.88 18.00 1.55

34.25 35.00 5.06

27.75 28.00 1.91

31.75 32.00 4.03

Inflation Mean Median St. dev. Foreign bank Mean Median St. dev. Banks concentration Mean Median St. dev. Shareholder rights Mean Median St. dev.

Corruption Mean Median St. dev.

3.29 3.10 0.47

4.66 4.70 0.65

5.68 5.70 0.05

4.97 5.09 0.38

3.09 3.05 0.43

4.09 3.80 0.45

4.70 4.40 0.75

3.12 3.15 0.33

3.78 3.80 0.15

Country credit rating Mean Median St. dev.

2.88 2.00 1.25

13.50 13.50 0.53

10.00 10.00 0.00

10.63 10.00 0.92

10.00 10.00 0.00

9.00 9.00 0.00

10.88 10.50 0.99

4.38 4.00 2.39

10.13 10.50 0.99

Number of obs.

60,567

36,297

22,393

21,551

11,984

3203

20,775

149,904

5902

297

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308 Table 1 (continued) 1995

1996

Panel B – across year Broad leverage Mean Median St. dev.

0.51 0.50 0.31

0.54 0.54 0.29

Narrow leverage Mean Median St. dev.

0.19 0.04 0.28

Tangibility Mean Median St. dev.

1998

1999

2000

2001

2002

0.58 0.59 0.31

0.60 0.62 0.31

0.64 0.67 0.32

0.66 0.69 0.32

0.67 0.69 0.33

0.70 0.72 0.32

0.16 0.00 0.26

0.16 0.00 0.27

0.18 0.00 0.28

0.20 0.00 0.30

0.21 0.01 0.31

0.22 0.03 0.31

0.24 0.08 0.32

0.41 0.40 0.28

0.38 0.36 0.26

0.39 0.37 0.26

0.40 0.38 0.26

0.39 0.37 0.26

0.37 0.35 0.26

0.38 0.36 0.26

0.38 0.36 0.25

Profitability Mean Median St. dev.

0.21 0.11 0.29

0.19 0.11 0.27

0.19 0.12 0.27

0.11 0.08 0.22

0.08 0.06 0.21

0.09 0.06 0.21

0.09 0.07 0.21

0.11 0.08 0.21

Total assets Mean Median St. dev.

6396 402 62,198

5175 242 68,355

4315 376 41,122

4305 377 40,658

4627 423 47,937

4763 443 51,468

6337 590 68,563

GDP growth Mean Median St. dev.

4.13 4.54 2.71

2.80 4.16 4.79

3.26 4.61 6.02

3.22 4.72 3.75

1.51 1.50 2.17

4.41 3.95 2.24

4.61 4.10 2.14

4.52 4.62 2.13

Inflation Mean Median St. dev.

29.24 28.30 15.78

31.53 23.05 35.10

143.23 10.58 346.48

15.44 10.63 16.98

9.42 3.30 14.11

11.06 9.80 13.57

8.68 5.74 9.98

5.16 3.32 6.75

Foreign bank Mean Median St. dev.

23.42 22.22 14.53

30.20 25.81 15.05

37.63 34.94 14.96

48.77 50.00 8.52

52.53 52.17 12.06

62.55 63.01 9.17

62.93 66.67 11.85

65.86 71.05 13.81

0.83 0.84 0.14

0.80 0.84 0.16

0.77 0.83 0.18

0.72 0.66 0.18

0.71 0.68 0.14

0.69 0.67 0.14

0.67 0.68 0.16

0.65 0.66 0.17

12.22 12.00 4.58

12.22 12.00 4.58

12.22 12.00 4.58

12.22 12.00 4.58

12.22 12.00 4.58

13.56 14.00 3.75

13.56 14.00 3.75

13.56 14.00 3.75

Minority shareholder rights Mean 10.00 Median 9.00 St. dev. 3.12

10.00 9.00 3.12

10.00 9.00 3.12

10.00 9.00 3.12

10.00 9.00 3.12

11.78 12.00 1.99

11.78 12.00 1.99

11.78 12.00 1.99

Corporate income tax Mean Median St. dev.

29.22 30.00 9.16

28.56 30.00 8.44

25.89 25.00 7.29

25.67 25.00 6.89

24.78 25.00 5.83

23.89 28.00 6.77

23.00 25.00 6.12

22.33 25.00 5.92

4.19 3.90 1.16

4.27 3.90 1.19

4.21 3.90 1.10

4.04 3.90 1.05

4.13 3.80 0.87

4.08 4.10 0.90

4.17 3.90 0.91

4.11 4.00 0.88

Country credit rating Mean Median St. dev.

9.22 10.00 3.27

9.22 10.00 3.27

9.22 10.00 3.27

9.00 10.00 3.50

8.44 10.00 3.84

8.89 10.00 3.62

9.00 10.00 3.71

9.33 10.00 3.32

Number of obs.

19,613

25,599

44,718

53,137

60,853

61,199

34,083

Banks concentration Mean Median St. dev. Shareholder rights Mean Median St. dev.

Corruption Mean Median St. dev.

1997

8244 324 769,610

33,374

Notes: Broad leverage is defined as total liabilities to total assets. Narrow leverage is defined as debt to debt plus shareholders’ funds. Tangibility is defined as tangible assets to total assets. Profitability is defined as before tax profit to total assets. Total assets are in thousands of USD. Foreign bank is defined as a share of foreign-owned banks assets to total banking sector assets. Bank concentration is defined as ratio of three biggest banks assets to the total banking sector assets. The shareholder rights and minority shareholder rights are indexes from Martynova and Renneboog (2010). Corruption is defined as corruption perceptions index.

298

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

have a small effect on leverage (no market timing). Still the macroeconomic conditions may affect the leverage through the fact that they proxy the growth opportunities in the overall economy. Since equity financing is less common in Eastern Europe the investment opportunities will be mostly financed by debt and therefore we would expect GDP growth to be positively related to leverage. As already noted, several empirical studies of capital structure have examined the relationship between leverage and a variety of largely firm-specific variables. Frank and Goyal (2009) note that seven variables—median industry leverage, market-to-book ratio, collateral, profit, dividend paying, logarithm of assets and expected inflation—perform best in explaining the leverage of US firms. In practice many of these variables will be fairly constant within a given industry at any given time. As a result a pure industry effect may capture much of the observed cross-sectional variation. Our main focus is to understand how much of observed variation of leverage cannot be accounted for by either industry effects or firm-specific measures and to propose the country characteristics that might do the job. The firm-specific characteristics used in this study follow Frank and Goyal (2009). Except, we have no information about dividend payments, and the market value of firms is available only for a small subsample of firms (only for a 20% of listed firms). Thus the two significant firm-specific characteristics based on Frank and Goyal (2009)—dividend payment and market-to-book ratio—are not included in the analysis. In addition firm age is included since both Nivorozhkin (2005) and De Haas and Peeters (2006) find it to be a significant determinant of leverage in transition economies. We define the firm-specific characteristics included in the analysis as follows: profitability = before-tax profit/total assets, tangibility = tangible fixed assets/total assets, log size = logarithm of total assets, and age dummies: established 1987–1995 has value one if the firm was established during early transition 1987–1995 and established after 1995 has value one if the firm was established after 1995 (variable definitions can be found in Appendix A). The current study augments the set of explanatory variables considered by Frank and Goyal (2009) by using a rich set of country-specific economic and institutional variables.4 We introduce nine country-specific variables into our analysis: inflation, GDP growth, corporate income tax rate, the share of foreign owned banks assets to total banking sector assets, the share of the three biggest banks assets to total banking sector assets, country credit rating, the corruption perception index, the shareholder rights protection index and the minority shareholder rights protection index. As has already been discussed GDP growth rate is included as a proxy for growth opportunities and the positive relationship with leverage is expected. GDP growth has been used in previous studies (see Frank and Goyal, 2009) and found to be positively related to leverage. Inflation and corporate tax rate indicate the tax shield benefit of debt and are expected to be also positively related to leverage. Empirical support for the prediction of positive relationship between the tax rate and the leverage has been weak. For example, several papers have found evidence that higher corporate taxes are negatively related to leverage (see Desai et al., 2004; Giannetti, 2003). The market share of foreign-owned banks is an important indicator in Eastern Europe because under socialism there was no competitive banking system, hence there was a lack of knowledge and experience of modern banking in the early 1990’s. A higher market share of foreign-owned banks reflects a higher quality of the banking system as well as larger competition among banks. This translates into more funds available in the domestic market and hence the higher leverage of firms.5 Note that the market share of foreign-owned banks is highly correlated (73%) with the ratio of foreign direct investment (FDI) to GDP. Hence, the higher market share of foreign-owned banks might be interpreted as a greater interest among foreign investors in general in a given economy. Another measure of banking sector competition is the concentration ratio of the three biggest banks assets to the assets of all banks. The higher the ratio the less competitive is the market and hence is expected to be negatively related to firm leverage. Additional institutional factors are included as alternative measures of the severity of asymmetric information. It is expected that the higher the corruption perception index (means lower corruption) the less severe is the asymmetric information problem. Hence, the positive relation between this index and external finance use is expected. Since the equity financing is less common in Eastern Europe we would expect positive relation between this index and leverage. Higher shareholder rights protection index and minority shareholder rights protection index should make the equity capital more appealing and hence could be negatively related to leverage. Though as argued before the equity financing is less used in Eastern Europe and therefore the effect could be also positive, proxying the over all decrease in asymmetric information and hence, boosting debt financing.6 Finally, country credit rating signals prudence and financial order in the country, hence higher the credit rating lower the asymmetric information problem and higher the leverage of firms. Previous capital structure studies based on firms from transition economies have generally focused on the level of leverage and its relation to firm-specific determinants. In a study of the early phase of transition, Cornelli et al. (1998) use data on Hungarian and Polish firms from the early 1990’s. They find that the level of leverage is lower than in developed economies and that the fraction of short-term financing dominates long-term debt. They estimate a static leverage regression, where the explanatory variables are tangibility, size, profitability and a dummy for state ownership. In contrast to studies on Western firms, these authors find that the share of tangible assets, which proxies the available collateral, is negatively related to

4 Note that Frank and Goyal (2009) experimented with several country-specific variables but all others besides inflation were less robust determinants of leverage. 5 See Giannetti and Ongena (2005,) for more details about the influence of foreign bank entry on domestic firm activities in Eastern Europe. 6 In addition improved creditor rights might increase debt financing but due to limited time-variation in the available creditor rights protection index it was left out from the empirical estimations.

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

299

leverage in the case of transition countries. They offer two explanations for this. First, they argue that through state ownership pre-transition firms effectively financed their fixed assets with equity and therefore the relationship to debt is negative. Second, the book value of fixed assets might differ from the market values. If there is a systematic bias it may be the case that high reported fixed assets are in fact overestimates of market values, such firms will have lower leverage than might otherwise be predicted. The authors thus report that Eastern European firms capital structure behaves differently from Western European firms with respect to firm-specific characteristics. It should be noted that this study is based on only two countries and covers a period when the transition process was just getting underway. The lack of country-specific variability in their study, however, means that they are unable to measure the significance of institutional and macroeconomic factors, which is the focus of the present paper. Later studies by Nivorozhkin (2005) and De Haas and Peeters (2006) explore the speed of adjustment of capital structure of firms in transition countries toward an hypothesized target level. In their framework actual leverage deviates from desired levels because of adjustment costs. Both papers use data from the Amadeus database available from Bureau Van Dijk, specifically the ‘‘top 200,000’’ sample which includes only relatively large firms. They adopt the methodology of Banerjee et al. (2004) which allows both the desired leverage and the adjustment speed to vary across firms and over time. De Haas and Peeters (2006) analyse ten countries over the period 1993–2001. Nivorozhkin (2005) analyses five countries over the period 1997–2001. Both papers find evidence that leverage levels were adjusting toward a target level determined by firm-specific factors. They conclude that deepening of Eastern European capital markets have enabled higher leverage levels for firms. It should be noted that their methodology presumes the existence of an adjustment process towards a target level. In contrast, Lemmon et al. (2008) have recently emphasized the strong persistence in observed differences in leverage across seemingly similar firms. They capture this persistence by using firm fixed effects. In this same spirit we will re-examine the adjustment of leverage in transition economies allowing for the possible persistence in leverage differences captured by country-specific (institutional) factors. There are several different leverage measures used in capital structure studies (see the discussion of leverage definitions in Rajan and Zingales, 1995). We use two leverage measures—broad leverage is defined as total liabilities over total book assets, while narrow leverage is defined as debt (both long-term and short-term debt) over the sum of debt and book value of the shareholder’ funds. The advantage of the former measure is that it is available for all firms in the data set; the shortcoming is that it is likely to overstate the true level of leverage. Since the theory of capital structure refers to the part of liabilities which are used for financing (in total liabilities some short-term items might be used for transactions only), the narrow leverage would seem to be a more relevant measure. But it is still possible that trade credit is used for financing as well and it would therefore be wrong to exclude it from a capital structure study. Leverage measures used in other studies on transition economies are similar to the ones used in this study. Cornelli et al. (1998) conduct their study on two measures, which corresponds to the two measures in this paper, and find that their results are robust to the leverage measures. Nivorozhkin (2005) uses a leverage measure which corresponds to the narrow leverage. De Haas and Peeters (2006) calculate the debt in leverage ratio as total liabilities minus trade credit. The measure they use is thus somewhere between the two measures used in this paper. 3. Data and methodology The firm-specific data used in this paper are taken from the Amadeus database available from Bureau Van Dijk. This database contains firm-level data from all over Europe. The Amadeus database is available in different sizes. Firms in this study are taken from the Amadeus Top 1 million companies.7 The analysis is based on eight years of data (1995–2002) for nine countries (Bulgaria, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, and Slovakia). The database consists not only of stock market-listed firms but, more importantly, also covers unlisted companies. Klapper et al. (2002) report that 86% of Eastern European firms in the Amadeus sample for 1999 have fewer than 250 employees. The data hence cover SMEs as well as large companies.8 The sample is unbalanced and the representativeness across countries varies.9 Romania has the greatest coverage. The largest firms are from Poland and the Czech and Slovak Republics while the smallest are from Bulgaria, Estonia and Romania (see Table 1 Panel A). The two leverage measures used in this study differ greatly from each other (see Table 1 Panel A and B). The average of broad leverage has increased from 52% to 69%, whereas the average of narrow leverage has increased from 19% to 24%. Both 7 For comparison the firms in De Haas and Peeters (2006) are from the Amadeus Top 200,000 companies, which covers fewer firms (smaller firms are not covered). 8 In our sample average firm has 160 employees (median 40 employees) and average total assets are $5.8 million (median $0.5 million). 9 For the Czech Republic and Estonia we know the size distribution of all firms across industries, which we compared with the size distribution of the Amadeus sample. The Amadeus sample over-estimates the share of largest firms, which is likely due to the sample selection criteria. Companies in the Amadeus Top 1 million sample have to meet at least one of the following criteria: operating revenue greater than 1 million euros, total assets greater than 2 million euros or number of employees above 10. In the robustness check I repeated the analysis on smaller sample of firms, where I randomly selected a number of observations from each country that corresponds to the relative size of that economy compared to the largest economy in my sample (hence in this subsample analysis I have close to 20,000 firms from Poland and 500 from Estonia (other countries are in between)). The results on this subsample are robust to the findings from whole sample. This subsample analysis shows that larger coverage of firms from some countries are not biasing the results. The results based on the subsample analysis are available on request.

300

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

measures express large variation across countries. Compared to firms from developed economies (see Rajan and Zingales, 1995) the leverage seems to be smaller for Eastern European firms. Smaller firm indebtedness in Eastern Europe might be explained by the fact that domestic credit provided by the banking sector (to GDP) is around 40% in Eastern European countries while it is more than 100% in many developed economies. The macroeconomic and institutional variables are obtained from a variety of sources. GDP growth and inflation are taken from World Bank World Development Indicators. Corporate tax rates are from Eurostat and KPMG. The market share of foreign-owned banks is from EBRD. The concentration ratio of the three biggest banks assets to the assets of all banks is taken from Beck et al. (2000). Country credit rating is adopted from S&P. The corruption perceptions index is from Institutional Investor.10 The shareholder rights and the minority shareholder rights protection indexes are from Martynova and Renneboog (2010).11 From Table 1 Panel B we can see that inflation has dropped during the sample period from close to 30% to as low as 5%. Also the market share of the three biggest banks has gone down from 83% to 65%. At the same time the share of foreign owned banks has gone up from 23% to 65%. Rest of the country-specific variables have varied less during the study period. To establish the main stylized facts about capital structure in transition countries we start with an analysis of variance (ANOVA) of leverage in three dimensions: industry, country and time. ANOVA is a special case of linear regression model where regressors are categorical variables. ANOVA allows to decompose the variation of leverage among different factors.12 The model can be written as,

Y ikjt ¼ a þ bj þ dk þ ct þ eijt ;

ð1Þ

where i, k, j, and t are the indexes of firm, industry, country, and year, respectively. Yikjt is the leverage of firm i, from industry k and country j in year t.13 bj is the country fixed effect, dk is the industry fixed effect14 and ct is the year effect. eijt is the random disturbance. We then explore what might account for the effects we have document using continuous explanatory variables introducing first the firm-specific variables widely used in past studies and then considering country-specific economic and institutional determinants. To compare our results to the preceding ANOVA, we can obtain a similar decomposition of total variation of the dependent variable in a linear model that combines categorical and continuous variables using analysis of covariance techniques, ANCOVA. The model controlling for firm-specific determinants can be written,

Y ikjt ¼ a þ bj þ dk þ ct þ fX ijt þ eijt ;

ð2Þ 15

where Xijt represents the three firm-specific variables — size, profitability and tangibility. The model that also controls for the time-varying country-specific factors can be written,

Y ikjt ¼ a þ bj þ dk þ ct þ fX ijt þ gC jt þ eijt ;

ð3Þ

where Cjt represent the country-specific time-variant variables. Next, we look at the coefficient estimates of the linear regression model to see the qualitative effect of leverage determinants. The regression analysis estimates similar specification to the Eqs. (2) and (3) with OLS.16 The minor difference is that instead of industry dummies we use the median industry leverage as a proxy for the industry effect. Also the firm age is controlled for in regression analyses.17 In addition, attempt to explain the change in leverage during the transition process is made, given the initial conditions. For each country we have defined a date corresponding to the end of transition.18 The estimated regression is in following form: 10 For country’s credit ratings and corruption perception index for some countries the early sample years have missing values. We have replaced the missing values with the earliest estimate available. 11 Martynova and Renneboog (2010) provide the values of the indexes only for years 1990, 1995, 2000, and 2005. In the current study for years 1995–1999 we keep the index constant at 1995 index value and for years 2000–2002 we keep the index constant at 2000 index value. 12 See Scheffe (1959) for a full discussion of ANOVA. 13 Leverage observations with extreme values (if leverage is less (more) than three times the inter quartile range away from 25th (75th) percentile) of leverage are excluded. Approximately 7000 firm years are lost due to this, this is less than 2% of total observations. 14 Firms from 51 industries are represented (NACE 2 digit classification). Firms from the financial intermediation sector are excluded from the study due to their specific liability structure. 15 We also experimented by adding firm age into the ANCOVA analysis; this did not change the pattern of results. 16 Results are robust for using fixed effects but since firm-specific fixed effects absorb the country dummies the OLS specification is preferred. 17 De Jong et al. (2008) argue in their capital structure study based on firms from 42 countries that the firm-specific determinants of leverage are also sensitive to the country factors, and this needs to be addressed. De Jong et al. (2008) rely on a two-step estimation procedure. In the first stage they estimate the firm leverage on firm-specific variables, country dummies and the interaction terms of country dummies with the firm-specific variables. In the second stage they take the coefficients from the first stage specification and use those as dependent variables in establishing the effect of country-specific factors (they will have 42 observation for each regression in the second stage). In current study we are unable to replicate their second stage estimation since we have only nine countries and hence we would not have enough data points for the analysis. 18 The date was defined based on EBRD transition indicators. EBRD has introduced 14 indicators scaling from 1 (bad) to 5 (good). For each year we have summed up the values of this indicators and we define the transition to be over when the sum gets above 42 (average of the indicators is above 3). Based on this for Hungary the transition end date would be 1995, which would not allow to do the before and after comparison, so the earliest possible date 1997 was used as an end of transition date. The end of transition date was assumed 1997 for Poland; 1998 for Czech Republic and Estonia; 1999 for Latvia; 2000 for Bulgaria, Lithuania and Romania; 2001 for Slovak Republic. The results are robust if a common transition end date is defined for all countries.

301

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

Y ikjA  Y ikjB ¼ a þ fX ijB þ gC jB þ eijt ;

ð4Þ

where the index A refers to the date one year after the end of the transition date and index B refers to the date one year before the transition end. This cross-sectional estimation allows to see what variables matters the most for the change in leverage. In a more parsimonious model we also include the before date leverage among the control variables. This allows to control for initial heterogeneity in capital structure. We perform the analysis separately on listed and unlisted firms. Since we consider being listed as a good signal for both domestic as well as foreign financiers, we expect that local institutions matter less for listed firms’ capital structure. 4. Results 4.1. What are the stylized facts about capital structure in transition countries? We present the results separately for listed and unlisted firms across the two leverage measures. Table 2 presents the results of the variance decomposition for listed firms. The number behind each variable name represents the fraction of partial sum of squares explained by a given variable. By construction each column sums up to 1. In Panel A the results for broad leverage are presented. In the 1st column country, industry and year dummies are included. Industry explains the most of the leverage variation and the adjusted R2 is 15%. In column (2) three time-varying firms-specific well known leverage determinants are added. The adjusted R2 goes up to 26%. Industry retains the largest explanatory power but tangibility explains also a large share of the leverage variation. The 3rd column is of interest since here both the country dummies as well as time-varying country variables are included. The adjusted R2 remains almost unchanged but the explanatory power of

Table 2 Variance decomposition for listed firms.

Panel A – broad leverage Country Year Industry Log size Profitability Tangibility GDP growth Inflation Foreign banks Bank concentration Shareholder rights Minority shareholder rights Corporate tax Corruption Country credit rating Adj. R2 Obs. Panel B – narrow leverage Country Year Industry Log size Profitability Tangibility GDP growth Inflation Foreign banks Bank concentration Shareholder rights Minority shareholder rights Corporate tax Corruption Country credit rating Adj. R2 Obs.

(1)

(2)

(3)

(4)

(5)

0.14 0.16 0.70

0.12 0.03 0.45 0.06 0.09 0.26

0.26 3513

0.03 0.47 0.06 0.11 0.25 0.02 0.00 0.00 0.01 0.00 0.00 0.01 0.00 0.03 0.26 3513

0.04 0.53 0.01 0.10 0.32

0.15 3513

0.05 0.01 0.48 0.08 0.10 0.25 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.27 3513

0.38 0.13 0.49

0.21 0.09 0.37 0.21 0.01 0.11

0.24 2909

0.06 0.44 0.23 0.02 0.14 0.00 0.01 0.06 0.00 0.01 0.00 0.02 0.00 0.00 0.23 2909

0.11 0.51 0.24 0.01 0.13

0.15 2909

0.12 0.02 0.43 0.25 0.01 0.12 0.00 0.01 0.02 0.00 0.00 0.00 0.01 0.00 0.01 0.25 2909

0.22 3513

0.20 2909

Notes: Numbers in cells refer to the fraction of partial sum of squares explained by given variable. Broad leverage is defined as total liabilities to total assets. Narrow leverage is defined as debt to debt plus shareholders’ funds. Industry is 2-digit NACE. Tangibility is defined as tangible assets to total assets. Profitability is defined as before tax profit to total assets. Foreign bank is defined as a share of foreign-owned banks assets to total banking sector assets. Bank concentration is defined as ratio of three biggest banks assets to the total banking sector assets. The shareholder rights and minority shareholder rights are indexes from Martynova and Renneboog (2010). Corruption is defined as corruption perceptions index.

302

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

country dummies decreases by half. Hence, we are able to explain a large share of leverage variation by known country macro/institutional variables. Still 48% of leverage variation is explained by industry dummies, followed by tangibility (25%), profitability (10%) and firms size (8%). The country dummies explain only 5% of leverage variation. Column (4) and (5) are included just for robustness to see how the model works when country factors are left out. Just leaving out the country dummies does not much affect the adjusted R2 confirming that country-specific variables do a good job in replacing country dummies. In the last column all country variables are left out and the adjusted R2 drops to 22%. Hence, we do have an evidence that country characteristics help to improve the explanatory power of the leverage estimation. In Panel B of Table 2 are the results for the narrow leverage. Differently from broad leverage the country dummies explain higher share of leverage variation. Roughly half of this explanatory power is wiped out when other country characteristics are included to the estimation. In full specification industry dummies explain 43% of leverage variation, followed by firm size with 25% and country dummies with 12%. Overall we can conclude that industry and firm-specific variables explain a bigger share of leverage variation than country variables for listed firms. In Table 3 the results for unlisted firms are reported. Country dummies explain a large share of narrow leverage variation. Looking at the complete specification in column (3) for narrow leverage measure the industry dummies still lead in explanatory power but country dummies are not far behind. From other explanatory variables the firm size stands out. For broad leverage profitability and tangibility do a good job in explaining leverage variation. Among country effects country credit rating has high explanatory power for both leverage measures. For narrow leverage also the market share of the foreign banks, shareholder rights protection index, corporate tax rate and the corruption perception index snatch up a substantial part of leverage variation. This indicates that different institutional factors can explain well the cross-country differences in leverage.

Table 3 Variance decomposition for unlisted firms.

Panel A – broad leverage Country Year Industry Log size Profitability Tangibility GDP growth Inflation Foreign banks Bank concentration Shareholder rights Minority shareholder rights Corporate tax Corruption Country credit rating Adj. R2 Obs. Panel B – narrow leverage Country Year Industry Log size Profitability Tangibility GDP growth Inflation Foreign banks Bank concentration Shareholder rights Minority shareholder rights Corporate tax Corruption Country credit rating Adj. R2 Obs.

(1)

(2)

(3)

(4)

(5)

0.34 0.24 0.42

0.17 0.06 0.08 0.02 0.28 0.40

0.23 379,575

0.01 0.11 0.03 0.29 0.48 0.00 0.01 0.00 0.03 0.00 0.02 0.00 0.01 0.01 0.22 379,575

0.07 0.12 0.08 0.26 0.47

0.10 379,575

0.10 0.00 0.09 0.03 0.31 0.46 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.24 379,575

0.79 0.04 0.18

0.62 0.03 0.20 0.15 0.00 0.00

0.11 330,659

0.13 330,659

0.27 0.02 0.34 0.24 0.00 0.01 0.02 0.01 0.01 0.02 0.01 0.01 0.00 0.00 0.05 0.13 330,659

0.12 0.25 0.21 0.01 0.00 0.03 0.01 0.05 0.00 0.06 0.01 0.09 0.07 0.10 0.12 330,659

0.20 379,575

0.05 0.34 0.60 0.01 0.00

0.06 330,659

Notes: Numbers in cells refer to the fraction of partial sum of squares explained by given variable. Broad leverage is defined as total liabilities to total assets. Narrow leverage is defined as debt to debt plus shareholders’ funds. Industry is 2-digit NACE. Tangibility is defined as tangible assets to total assets. Profitability is defined as before tax profit to total assets. Foreign bank is defined as a share of foreign-owned banks assets to total banking sector assets. Bank concentration is defined as ratio of three biggest banks assets to the total banking sector assets. The shareholder rights and minority shareholder rights are indexes from Martynova and Renneboog (2010). Corruption is defined as corruption perceptions index.

303

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308 Table 4 Variance decomposition for listed firms by size class. Source

Broad leverage

Country Year Industry Adj. R2 Obs.

Narrow leverage

Size < 4

Size 4

Size 5

Size < 4

Size 4

Size 5

0.03 0.21 0.76 0.22 1160

0.29 0.10 0.62 0.22 1650

0.21 0.06 0.73 0.35 703

0.04 0.38 0.58 0.16 1107

0.29 0.08 0.62 0.18 1284

0.38 0.03 0.59 0.35 518

Notes: Numbers in cells refer to the fraction of partial sum of squares explained by given variable. Broad leverage is defined as total liabilities to total assets. Narrow leverage is defined as debt to debt plus shareholders’ funds. Industry is 2-digit NACE. Firm size classes: Class < four total assets (TA) smaller $5 million, Class 4 TA between $5 and 50 million, and Class 5 TA above $50 million.

Table 5 Variance decomposition for unlisted firms by size class. Broad leverage

Country Year Industry Adj. R2 Obs.

Narrow leverage

Size 1

Size 2

Size 3

Size 4

Size 5

Size 1

Size 2

Size 3

Size 4

Size 5

0.48 0.27 0.25 0.10 242,186

0.12 0.28 0.61 0.09 42,272

0.10 0.20 0.71 0.10 46,710

0.03 0.11 0.86 0.15 43,425

0.05 0.06 0.89 0.25 4982

0.65 0.09 0.26 0.08 220,014

0.74 0.05 0.20 0.11 35,183

0.78 0.02 0.20 0.12 37,282

0.66 0.05 0.29 0.14 34,260

0.44 0.06 0.50 0.22 3920

Notes: Numbers in cells refer to the fraction of partial sum of squares explained by given variable. Broad leverage is defined as total liabilities to total assets. Narrow leverage is defined as debt to debt plus shareholders’ funds. Industry is 2-digit NACE. Firm size classes: Class 1 total assets (TA) smaller than $1 million, Class 2 TA between $1 and 2 million, Class 3 TA between $2 and 5 million, Class 4 TA between $5 and 50 million, and Class 5 TA above $50 million.

Comparing the results across listed and unlisted firms we can conclude that the country characteristics explain higher share of leverage variation for unlisted firms. Also looking at the adjusted R2 in column (3) and (5) for narrow leverage we see that the marginal drop in the adjusted R2 is larger for the unlisted than for listed firms, when the country characteristics are left out. The importance of country dummies shows that the time-invariant part of leverage is important. Similar high persistence in leverage is pointed out by Lemmon et al. (2008) who show that firm-specific fixed effect soak up most of the leverage variation. Neither time fixed effects or time-varying firm-specific variables are able to decrease the importance of them. In this study we avoid using firm fixed effects because they embody the country effects and would not let to evaluate the latter. The different results obtained for listed and unlisted firms could be explained by the fact that listed firms are larger.19 To see whether the results differ due to size differences we conducted variance decomposition across five size classes.20 Table 4 presents the results for listed firms. For both leverage measures, industry factors explain the most for all size classes.21 For unlisted firms (Table 5) the results are different for firms from different size classes. Country factors explain the most for the smallest firms’ broad leverage variation. For firms from the four larger size classes, industry factors dominate in explaining the broad leverage variation. This is consistent with the hypothesis that smaller firms rely more heavily on the local financial market. For unlisted firms’ narrow leverage, country factors explain the most for the four smallest size classes. The explanatory share of industry factors becomes dominant for the largest firms. These results on size classes confirm that for the smallest unlisted firms, country factors are the most significant leverage determinants for both leverage measures.22 Those firms are clearly more constrained by the local financial market than are other firms. For unlisted firms variance decomposition results differ for different leverage measures, so it is really important which leverage measure is used. The main difference between the two measures comes from the current liability side—narrow leverage takes into account only short-term debt (not all short-term liabilities). Narrow leverage captures the loan capacity of the firm, which seems to be highly country-specific for unlisted firms.23 Broad leverage, on the other hand, captures nondebt liabilities like trade credit, which is a particularly important source of funds for more financially constrained firms (see Petersen and Rajan, 1997). The trade credit is also a more important source of funds for Eastern European firms than for Western European firms. Trade credit is 43% of total liabilities in our Eastern European sample while it is only 24% in Jõeveer (2005) 19 The difference might also be caused by ownership—stock market-listed companies are more likely to have foreign owners, which might ease their financing needs. As we do not have information about foreign ownership we cannot test this. 20 The size classes are following: total assets up to $1 million, between $1 and 2 million, between $2 and 5 million, between $5 and 50 million and above $50 million. 21 We combined the three smallest size classes due to lack of observations. 22 A different size classification does not change the findings. 23 Many firms have no debt at all and hence the narrow leverage variable has many zeros.

304

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

Table 6 Leverage regression in 1995–2002. Broad leverage Panel A – Listed firms Const. Tangibility Profitability Log total assets Established 1987–1995 Established after 1995 Median industry leverage

.072 (.054) .268 (.037)⁄⁄⁄ .289 (.055)⁄⁄⁄ .025 (.005)⁄⁄⁄ .043 (.023)⁄ .012 (.023) .802 (.057)⁄⁄⁄

GDP growth Inflation Foreign banks Banks concentration Shareholder rights Minority shareholder rights Corporate income tax Corruption Country credit rating Year Country R2 Obs. Panel B – Unlisted firms Const. Tangibility Profitability Log total assets Established 1987–1995 Established after 1995 Median industry leverage GDP growth Inflation Foreign banks Banks concentration Shareholder rights Minority shareholder rights Corporate income tax Corruption

Yes Yes .26 3513 .602 (.006)⁄⁄⁄ .405 (.003)⁄⁄⁄ .335 (.004)⁄⁄⁄ .011 (.0006)⁄⁄⁄ .06 (.004)⁄⁄⁄ .102 (.004)⁄⁄⁄ .091 (.006)⁄⁄⁄

Narrow leverage .084 (.118) .258 (.038)⁄⁄⁄ .303 (.057)⁄⁄⁄ .027 (.005)⁄⁄⁄ .046 (.023)⁄⁄ .01 (.023) .791 (.058)⁄⁄⁄ .007 (.001)⁄⁄⁄ .00003 (.00002)⁄⁄ .0006 (.0006) .061 (.087) .00009 (.008) .006 (.009) .0008 (.002) .038 (.011)⁄⁄⁄ .009 (.005)⁄⁄ Yes Yes .268 3513

.118 (.113) .256 (.037)⁄⁄⁄ .316 (.057)⁄⁄⁄ .023 (.005)⁄⁄⁄ .022 (.02) .019 (.023) .79 (.057)⁄⁄⁄ .008 (.002)⁄⁄⁄ .00004 (.00002)⁄⁄ .0002 (.0007) .185 (.082)⁄⁄ .005 (.004) .003 (.005) .003 (.002)⁄⁄ .023 (.011)⁄⁄ .013 (.003)⁄⁄⁄ Yes No .256 3513

.082 (.047)⁄ .133 (.026)⁄⁄⁄ .077 (.037)⁄⁄ .03 (.005)⁄⁄⁄ .024 (.02) .012 (.018) .688 (.071)⁄⁄⁄

.587 (.017)⁄⁄⁄ .403 (.003)⁄⁄⁄ .331 (.004)⁄⁄⁄ .011 (.0006)⁄⁄⁄ .068 (.004)⁄⁄⁄ .109 (.004)⁄⁄⁄ .092 (.006)⁄⁄⁄ 0 (.0002)⁄⁄⁄ .00002 (3.96e06)⁄⁄⁄ .0005 (.00008)⁄⁄⁄ .045 (.014)⁄⁄⁄ .015 (.001)⁄⁄⁄ .024 (.001)⁄⁄⁄ .0006 (.0003)⁄⁄ .014 (.002)⁄⁄⁄

.407 (.013)⁄⁄⁄ .408 (.003)⁄⁄⁄ .321 (.004)⁄⁄⁄ .013 (.0005)⁄⁄⁄ .1 (.004)⁄⁄⁄ .114 (.004)⁄⁄⁄ .097 (.006)⁄⁄⁄ .0003 (.0002)⁄ 0 (4.03e06)⁄⁄⁄ .0007 (.0001)⁄⁄⁄ .25 (.008)⁄⁄⁄ .003 (.0004)⁄⁄⁄ .017 (.0006)⁄⁄⁄ .003 (.0002)⁄⁄⁄ .025 (.001)⁄⁄⁄

.134 (.006)⁄⁄⁄ .014 (.003)⁄⁄⁄ .031 (.003)⁄⁄⁄ .023 (.0005)⁄⁄⁄ .09 (.004)⁄⁄⁄ .095 (.004)⁄⁄⁄ .151 (.008)⁄⁄⁄

Yes Yes .248 2909

.079 (.138) .127 (.027)⁄⁄⁄ .092 (.04)⁄⁄ .031 (.005)⁄⁄⁄ .025 (.02) .011 (.018) .669 (.072)⁄⁄⁄ .0005 (.002) .00006 (.00002)⁄⁄⁄ .001 (.0006)⁄⁄⁄ .076 (.09) .01 (.008) .012 (.007)⁄ .003 (.002)⁄⁄ .007 (.012) .013 (.005)⁄⁄ Yes Yes .26 2909

.108 (.113) .133 (.026)⁄⁄⁄ .109 (.038)⁄⁄⁄ .026 (.005)⁄⁄⁄ .006 (.017) .025 (.018) .671 (.073)⁄⁄⁄ .0006 (.002) .00006 (.00002)⁄⁄⁄ .002 (.0007)⁄⁄⁄ .094 (.075) .01 (.004)⁄⁄ .001 (.006) .004 (.001)⁄⁄⁄ .005 (.011) .001 (.004) Yes No .239 2909

.193 (.021)⁄⁄⁄ .014 (.003)⁄⁄⁄ .029 (.003)⁄⁄⁄ .023 (.0005)⁄⁄⁄ .097 (.004)⁄⁄⁄ .101 (.004)⁄⁄⁄ .149 (.008)⁄⁄⁄ .004 (.0002)⁄⁄⁄ .00006 (3.61e06)⁄⁄⁄ 0 (.0001)⁄⁄⁄ .329 (.017)⁄⁄⁄ .027 (.002)⁄⁄⁄ .027 (.002)⁄⁄⁄ .0007 (.0003)⁄⁄ .003 (.002)

.256 (.014)⁄⁄⁄ .01 (.003)⁄⁄⁄ .034 (.003)⁄⁄⁄ .025 (.0005)⁄⁄⁄ .097 (.004)⁄⁄⁄ .093 (.004)⁄⁄⁄ .15 (.008)⁄⁄⁄ .006 (.0002)⁄⁄⁄ .00007 (3.63e06)⁄⁄⁄ .003 (.0001)⁄⁄⁄ .02 (.008)⁄⁄ .013 (.0005)⁄⁄⁄ .007 (.0006)⁄⁄⁄ .008 (.0002)⁄⁄⁄ .045 (.001)⁄⁄⁄

305

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308 Table 6 (continued) Broad leverage Country credit rating Year Country R2 Obs.

Yes Yes .225 373,980

Narrow leverage .018 (.0007)⁄⁄⁄ Yes Yes .229 373,980

.005 (.0004)⁄⁄⁄ Yes No .208 373,980

Yes Yes .112 324,624

.024 (.0008)⁄⁄⁄ Yes Yes .117 324,624

.014 (.0004)⁄⁄⁄ Yes No .101 324,624

Notes: Broad leverage is defined as total liabilities to total assets. Narrow leverage is defined as debt to debt plus shareholders’ funds. Tangibility is defined as tangible assets to total assets. Profitability is defined as before tax profit to total assets. Established 1987–1995 is a dummy equal to one if the firm was established between 1987 and 1995. Established after 1995 is a dummy equal to one if the firm was established after 1995. Foreign bank is defined as a share of foreign-owned banks assets to total banking sector assets. Bank concentration is defined as ratio of three biggest banks assets to the total banking sector assets. The shareholder rights and minority shareholder rights are indexes from Martynova and Renneboog (2010). Corruption is defined as corruption perceptions index. Year refers to year fixed effects. Country refers to country fixed effects. Standard errors are in brackets. Standard errors are based on clustering across firms. ⁄ Denote significance at the 10% levels. ⁄⁄ Denote significance at the 5% levels. ⁄⁄⁄ Denote significance at the 1% levels.

Table 7 Change in broad leverage (after–before transition) regression. Listed firms Const.

.319 (.627)

Broad leverage B Tangibility B Profitability B Log total assets B Established 1987–1995 Established after 1995 Median industry leverage B GDP growth B Inflation B Foreign banks B Banks concentration B Shareholder rights B Minority shareholder rights B Corporate income tax B Corruption B Country credit rating B R2 Obs.

.078 (.044)⁄ .346 (.09)⁄⁄⁄ .004 (.007) .008 (.025) .049 (.029)⁄ .015 (.083) .004 (.009) .0004 (.0001)⁄⁄⁄ .004 (.004) .233 (.34) .018 (.01)⁄ .032 (.017)⁄ .007 (.008) .018 (.043) .02 (.016) .189 414

Unlisted firms .181 (.596) .162 (.057)⁄⁄⁄ .037 (.044) .42 (.091)⁄⁄⁄ .007 (.007) .007 (.024) .038 (.028) .108 (.095) .005 (.008) .0004 (.0001)⁄⁄⁄ .003 (.004) .279 (.325) .017 (.01)⁄ .033 (.016)⁄⁄ .008 (.007) .019 (.04) .017 (.016) .224 414

.131 (.089)

.121 (.008)⁄⁄⁄ .089 (.016)⁄⁄⁄ .002 (.002)⁄⁄ .029 (.005)⁄⁄⁄ .028 (.006)⁄⁄⁄ .048 (.011)⁄⁄⁄ .002 (.002) .00003 (.00005) .001 (.0006) .151 (.053)⁄⁄⁄ .004 (.002)⁄⁄ .001 (.003) .001 (.001) .007 (.005) .001 (.002) .019 41,382

.486 (.101)⁄⁄⁄ .523 (.069)⁄⁄⁄ .107 (.029)⁄⁄⁄ .102 (.022)⁄⁄⁄ .01 (.001)⁄⁄⁄ .027 (.009)⁄⁄⁄ .028 (.01)⁄⁄⁄ .008 (.012) .003 (.002)⁄⁄ .00008 (.00004)⁄⁄ 0 (.0006)⁄⁄ .056 (.043) .001 (.001) .001 (.002) .007 (.001)⁄⁄⁄ .041 (.007)⁄⁄⁄ .002 (.002)⁄⁄ .332 41,382

Notes: Broad leverage is defined as total liabilities to total assets. Standard errors are based on clustering across firms. Tangibility is defined as tangible assets to total assets. Profitability is defined as before tax profit to total assets. Established 1987–1995 is a dummy equal to one if the firm was established between 1987 and 1995. Established after 1995 is a dummy equal to one if the firm was established after 1995. Foreign bank is defined as a share of foreignowned banks assets to total banking sector assets. Bank concentration is defined as ratio of three biggest banks assets to the total banking sector assets. The shareholder rights and minority shareholder rights are indexes from Martynova and Renneboog (2010). Corruption is defined as corruption perceptions index. Before date is 1996 for Hungary and Poland, 1997 for Czech Republic and Estonia, 1998 for Latvia, 1999 for Bulgaria, Lithuania and Romania, 2000 for Slovakia. Standard errors are in brackets. ⁄ Denote significance at the 10% levels. ⁄⁄ Denote significance at the 5% levels. ⁄⁄⁄ Denote significance at the 1% levels.

306

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

sample of firms from ten Western European countries. The non-debt liabilities in Eastern Europe might have been used as substitutes for debt (if the latter was not available) so that the country-specific variation in broad leverage is evaporated. The variance decomposition results are comparable to the findings from Western European countries presented in Jõeveer (2005). For the listed firms in Western Europe, industry factors were the most significant determinants of leverage variation irrespective of firm size. For unlisted Western firms the country factors always explained the largest share of narrow leverage variation irrespective of size. For unlisted Western firms’ broad leverage, country factors mattered the most for the four smaller size classes while for the largest size class, industry factors turned out to be the most significant. The average firm in Western Europe is larger than in Eastern Europe, which might explain why we observe a change in the explanatory power of country and industry factors in smaller size classes in the Eastern European sample. The firms in size classes 2–4 are relatively larger than the average firm in the Eastern European sample than in the Western European sample. Hence, the variance decomposition stresses the importance of country factors for small unlisted companies’ leverage variation and it is irrelevant whether those firms are drawn from the pool of developed or less developed economies. The financing mix of both Eastern and Western European small firms is, compared to large firms, less dependent on firm-specific technological factors and more dependent on country of incorporation factors. A comparison of the results of Eastern and Western European firm leverage analyses does not support the initial expectation that the lesser-developed financial markets in the East might cause country factors to be more pronounced for the firm’s capital structure choice. 4.2. What accounts for cross-country differences in capital structure during the transition process? The preceding results have established important cross-country differences in capital structure among transition countries even after controlling for a full range of industry and firm-specific factors. We have also seen that country-specific economic and institutional factors account for some but not all of the pure country effects. We now consider in more detail these country-specific variables to see which appear most important and also whether initial firm-specific and country-specific factors are correlated with observed changes of capital structure during the transition process. The OLS estimates (with robust standard errors) of the regression analysis are presented in Tables 6 and 7.24 The countryspecific macroeconomic and institutional factors are included in columns (2), (3), (5), and (6). The significance and the direction of the effect of country-specific factors vary across leverage measures. In columns (3) and (6) the country fixed effects are left out and this also changes the significance of some country-specific time-varying variables confirming the idea that country dummies soak up the effect of those variables. For explaining the effect of country-specific variables on leverage we focus on columns (2) and (5). The results of listed firms leverage regression are reported in Table 6 Panel A. All coefficients on firm-specific variables have expected signs except the one on tangibility, which has a negative and statistically significant sign. This confirms the results of previous studies on transition countries (Cornelli et al., 1998) but contradicts the predictions of theoretical studies and empirical findings from developed countries (Rajan and Zingales, 1995). The inflation and country credit rating have negative signs in both leverage regressions contradicting the predictions. In addition for broad leverage we observe a statistically significant negative effect on GDP growth (contradicting the prediction) and a statistically positive effect on the corruption perception index (less corruption higher leverage, confirming the prediction). For narrow leverage we observe negative significant coefficients for corporate tax and the foreign banks concentration, which contradicts theoretical arguments. For unlisted firms (Table 6 Panel B) the results vary quite a bit across leverage definitions. The coefficient on profitability is negative and statistically significant meaning that the more profitable firms are likely to have less debt, this is in line with many empirical studies (Frank and Goyal, 2009). The logarithm of size and tangibility enter with negative signs in the broad leverage regression and with a positive signs in the narrow leverage regression. This finding stresses once again that for unlisted firms the two leverage ratios measure different things. Negative sign for size variable is also found in capital structure papers based on SMEs (Van Der Wijst and Thurik, 1993; Hall et al., 2004; Sogorb-Mira, 2005), where leverage is defined as the short-term debt to total debt. Based on age dummies included in the regression, the younger firms are shown to be more leveraged than older firms. Hence, we do not observe that an established reputation would lead to higher leverage as expected. One explanation for this might be that older firms have enough internal funds and do not need debt finance. Compared to listed firms the country factors have more significant coefficients. The positive coefficients on GDP growth, foreign bank concentration, and corporate tax; and negative coefficient on shareholder rights protection index confirm the theoretical hypothesis. While positive coefficients on banks concentration and minority shareholder rights protection index; and negative coefficients on inflation and country credit rating contradict the expectations. In addition for broad leverage the corruption perception index has a positive significant coefficient supporting our hypothesis. The final estimation is looking for factors explaining the change in leverage during the transition process. The results are reported in Table 7. The results are provided only for broad leverage measure.25 The initial profitability has negative significant coefficient indicating that initially highly profitably firms decrease their leverage during the transition process. In column 24

The OLS estimation results presented here are very similar to the results achieved by fixed effects (available on request). This analyses demands firms to have three consecutive years of data in the data set, which decreases the sample substantially and limits the use of narrow leverage. 25

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

307

(2) and (4) also the initial leverage is included. This expanded specification controls for initial heterogeneity among firm leverage. The initial leverage enters with a negative coefficient. This means that firms with initially high leverage decreased their leverage during the transition process while firms with initially low leverage increased their leverage. This effect is much larger for unlisted firms. The inclusion of initial leverage improves the estimation fit especially for unlisted firms. Several initial country factors are statistically significant and help to explain the change in leverage. More initial country factors affect the leverage change for unlisted compared to listed firms reinforcing the importance of later for the development of capital structure of the unlisted firms. 5. Conclusions In this paper we study the importance of firm-specific, country institutional, and macroeconomic factors for determining the capital structure of firms. The analysis is based on firm-level data from nine Eastern European countries in 1995–2002. We use two measures of leverage in this paper. We find that the largest share of listed firms’ leverage variation (irrespective of leverage measure or size category) is explained by industry factors. For unlisted firms, in contrast, the results are not robust to the leverage measure used. For broad leverage the firm-specific factors explain the most while for narrow leverage the country-specific factors dominate. Further, the unmeasurable country institutional differences explain as much as 25% of unlisted firms broad leverage variation. The results across size classes show that for smaller unlisted firms, country factors are the most significant explanatory factors for both leverage measures. Smaller firms seem to be more constrained by the financial market in their country of incorporation. The regression analysis of leverage finds some surprising coefficients for some firm-specific variables in some specifications (negative signs on tangibility and firm size). More importantly we find that country characteristics are significant determinants of leverage and that especially for unlisted firms. Based on our results a policy maker may derive some policy implications. For example if the policy maker prefers lower firm leverage (because the higher leveraged firms are more likely to face financial distress) she could support the decrease in the banking market concentration or try cutting the corporate tax rates. Also improving the shareholder rights may lead to the lower firm leverage. The findings of this study stress the importance of country institutional variables on the capital structure of firms. Acknowledgments This paper was written during the author’s stay at the Research Department of Bank of Estonia. The data work on this project started during the author’s research visit at the William Davidson Institute (WDI), University of Michigan Business School. The hospitality and support of these institutions is gratefully acknowledged. Ronald Anderson and Šteˇpán Jurajda provided many valuable discussions and suggestions. Finally, the Editor and the two anonymous referees are thanked for their helpful comments. Appendix A. Variable definitions Broad leverage Narrow leverage Tangibility Profitability Established 1987–1995 Established after 1995 Foreign bank Bank concentration

Total liabilities to total book assets Long-term debt plus short-term debt to long-term debt plus short-term debt plus book value of shareholders’ funds Tangible assets to total book assets Before tax profit to total book assets Dummy equal to one if the firm was established between 1987-1995 Dummy equal to one if the firm was established after 1995 Share of foreign-owned banks assets to total banking sector assets Ratio of three biggest banks assets to the total banking sector assets

References Baker, M., Wurgler, J., 2002. Market timing and capital structure. Journal of Finance 57, 1–32. Bancel, F., Mittoo, U.R., 2004. Cross-country determinants of capital structure choice: a survey of European firms. Financial Management 33, 103–132. Banerjee, S., Heshmati, A., Wihlborg, C., 2004. The dynamics of capital structure. Monetary Integration, Markets and Regulation 4, 274–297. Beck, T., Demirgüç-Kunt, A., Levine, R., 2000. A new database on financial development and structure. World Bank Economic Review, 597–605. Berglof, E., Bolton, P., 2002. The great divide and beyond financial architecture in transition. Journal of Economic Perspectives 16, 77–100. Booth, L., Aivazian, V., Demirgüç-Kunt, A., Maksimovic, V., 2001. Capital structure in developing countries. Journal of Finance 56, 87–130. Cornelli, F., Portes, R., and Schaffer, M.E., 1998, Financial Structure of Firms in the CEECs. In: Olivier, Bouin, Fabrizio, Coricelli, Françoise Lemoine (Eds.), Different Paths to a Market Economy: China and European Economies in Transition, CEPR/CEPII/OECD. Decoure, N., 2007. The determinants of capital structure in transition economies. International Review of Economics and Finance 16, 400–415. De Haas, R., Peeters, M., 2006. The dynamic adjustment towards target capital structures of firms in transition economies. Economics of Transition 14, 133– 169.

308

K. Jõeveer / Journal of Comparative Economics 41 (2013) 294–308

De Jong, A., Kabir, R., Nguyen, T.T., 2008. Capital structure around the world: the roles of firm- and country-specific determinants. Journal of Banking and Finance 32, 1955–1969. Desai, M.A., Foley, C.F., Hines, J.R., 2004. A multinational perspective on capital structure choice and international capital markets. Journal of Finance 59, 2451–2487. Frank, M.Z., Goyal, V.K., 2009. Capital structure decisions: which factors are reliably important? Financial Management 38, 1–37. Giannetti, M., 2003. Do better institutions mitigate agency problems? Evidence from corporate finance choices. Journal of Financial and Quantitative Analysis 38, 185–212. Giannetti, M., Ongena, S., 2005. Financial Integration and Entrepreneurial Activity: Evidence from Foreign Bank Entry in Emerging Markets. CEPR Working Paper, 5151. Hall, G.C., Hutchinson, P.J., Michaelas, N., 2004. Determinats of the capital structure of European SMEs. Journal of Business Finance and Accounting 31, 711– 728. Jõeveer, K., 2005, What do we know about the capital structure of small firms? CERGE-EI Working Paper, 283. Klapper, L.F., Sarria-Allende, V., Sulla, V., 2002. Small- and medium-size enterprise financing in eastern Europe. WB Policy Research Working Paper, 2933. Lemmon, M.L., Roberts, M.R., Zender, J.F., 2008. Back to the beginning: persistence and the cross-section of corporate capital structure. Journal of Finance 58, 1575–1608. Martynova, M., Renneboog, L., 2010. A Corporate Governance Index: Convergence and Diversity of National Corporate Governance Regulations. CentER Discussion Paper. Miller, M.H., 1977. Debt and taxes. Journal of Finance 32, 261–275. Modigliani, F., Miller, M., 1958. The cost of capital, corporation finance and the theory of investment. The American Economic Review 48, 261–297. Myers, S.C., 1984. The capital structure puzzle. Journal of Finance 39, 575–592. Nivorozhkin, E., 2005. Financing choices of firms in EU accession countries. Emerging Market Review 6, 138–169. Petersen, M.A., Rajan, R.G., 1997. Trade credit: theories and evidence. Review of Financial Studies 10, 160–177. Psillaki, M., Daskalakis, N., 2009. Are the determinants of capital structure country or firm specific? Small Business Economics 33, 319–333. Rajan, R., Zingales, L., 1995. What do we know about capital structure? Some evidence from international data. Journal of Finance 50, 1421–1460. Rubinstein, M., 2003. Great moments in financial economics: II. Modigliani–Miller theorem. Journal of Investment Management 1, 7–13. Scheffe, H., 1959. The Analyses of Variance. John Wiley & Sons. Schmukler, S., Vesperoni, E., 2001. Firms’ financing choices in bank-based and market-based economies. In: Asli, Demirgüç-Kunt, Ross, Levine (Eds.), Financial Structure and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development. MIT Press. Sogorb-Mira, F., 2005. How SME uniqueness affects capital structure: evidence from a 1994–1998 spanish data panel. Small Business Economics 25, 447– 457. Van Der Wijst, N., Thurik, R., 1993. Determinants of small firm debt ratios: an analysis of retail panel data. Small Business Economics 5, 55–65. Williams, J.B., 1938. The Theory of Investment Value. Harvard University Press.