Impact of the financial crisis on firms’ capital structure in
Greece, Italy, The Netherlands and Sweden
By I. Guftometros
Supervisor: Dr. H. Vrolijk
Co-assessor: Dr. W. Westerman
International Financial Management
Faculty of Economics and Business
University of Groningen
January 2015
2
Abstract
This study examines the impact of the financial crisis on the capital structure decision of 412 Greek, Italian, Dutch and Swedish firms. The results show that leverage ratios increase from the pre-crisis (2005-2007) to the post-crisis (2009-2011) period. Using balanced panel data this study further shows that the determinants of leverage have changed as well after the crisis. Overall, tangibility of assets and liquidity of assets become more important whereas the change of other determinants differs among countries. Furthermore, some light is shed on the impact the crisis had on the country-specific determinants of capital structure.
JEL Classification: G32
Keywords: Capital structure, leverage, financial crisis, Greece, Italy, The Netherlands, Sweden
1. Introduction
In pursuit of maximizing firm value financial managers must establish an optimal capital structure. Prior studies show that firms’ capital structure, or leverage ratio, is not only influenced by firm-specific factors but by the institutional environment and macro-economic conditions as well (Booth et al., 2001; De Jong et al., 2008). However, little has been written about how disruptions (e.g. major shocks) affect capital structure decisions. The recent financial crisis provides an opportunity to do so. This study examines the impact the recent financial crisis had on the capital structure of 412 listed European firms coming from Greece, Italy, The Netherlands and Sweden. The selection of these countries is based on a number of reasons. Many studies regarding the determinants of capital structure include either a large number of countries or a single country, which is inconvenient for an analysis on the differences between countries. Also, current capital structure literature predominately focuses on U.S. firms and large European nations such as the U.K., France and Germany (Antoniou and Paudyal, 2002). In-depth analysis of these four smaller European countries could extant our understanding of the role the institutional environment has in capital structure decisions. Second, even though all part of the EU, these countries have a very different institutional environment to begin with. Greece, Italy and The Netherlands are part of the European Monetary Union (EMU) but Sweden is not. Sweden is also the only country which financial orientation is market-based whereas the other countries have a bank-oriented economy1. Moreover, all countries fared very differently under the recent financial crisis. Greece and Italy were particularly hit hard by the financial crisis and
1
3
both countries even ended up in a sovereign debt crisis. These institutional differences may affect firms’ their capital structure decisions. Consequently the aim of this study is to answer the following research question;
How did the financial crisis impact the determinants of capital structure of Greek, Italian, Dutch and Swedish firms?
This study shows that leverage ratios in the sample countries have changed after the recent crisis. The change in leverage ratio differed per country and depends on how leverage is measured. Another major contribution is that not only the leverage ratios changed but their determinants are as well affected by the crisis. For example, the effect of assets tangibility increased after the crisis which suggests that collateral became more important. The same effect was found for liquidity. This study also aims to explain how differences in the institutional and macro-economic environment of firms can affect their capital structure decisions. The results are not definite but suggest that country-specific determinants of capital structure are also affected by the crisis. This should be a concern for future research.
1.1. Financial crisis
4
The next chapter discusses the relevant theory on capital structure. Chapter 3 explains the methodology that is used in this paper. Chapter 4 presents the results. The discussion of the results is done in chapter 5. The final chapter provides concluding remarks and directions for future research.
2. Theoretical framework
This chapter is divided in four parts. First, theory on capital structure and leverage is discussed. In the second part the firm-specific determinants of capital structure is discussed followed by an analysis of the country-specific determinants in the third part. The final part summarizes prior studies on the impact of crises on capital structure.
2.1. Leverage
Since the introduction of the theory on capital structure by Modigliani and Miller (1958) the subject has been widely studied. Capital structure theory attempts to explain what mix of debt and equity firms use to finance their investments. Commonly capital structure is researched by analyzing the leverage ratio of firms (Booth et al., 2001; De Jong et al., 2008; Rajan and Zingales, 1995). However, literature has proposed different ways to measure leverage. First, leverage can be calculated by taking either market values or book values of equity. Book values are based on accounting numbers which represents historic values. Book values are easier to calculate but do not account for the change in value (appreciation or depreciation) of assets. Older studies tend to include only book values while newer studies often use market values or, as robustness check, include both. Since this study focuses on large listed firms the use of actual, and not historical, figures in the analysis of financing decisions is of importance. Therefore only market values of leverage are analyzed. Furthermore, the type of debt that is included can have a huge impact on how to interpret the results. Some studies only analyze the total debt leverage (Deesomsak et al., 2004) while others also include long-term and short-term leverage in their analysis (Booth et al., 2001; De Jong et al., 2008). Following De Jong et al. (2008), among others, I mainly use the long-term debt leverage ratio instead of the total debt leverage ratio2. Short-term debt largely consists of trade credit and is therefore excluded. Trade credit is under the influence of completely other determinants and makes examination of the results difficult to interpret (De Jong et al., 2008). However, since short-term credit, and the availability thereof, is heavily
2
5
impacted by the crisis this study does check for the results of short-term leverage and presents some results as a side note or in the appendix.
2.2. Firm-specific determinants of capital structure
The following firm-specific variables have been found significant in earlier research and are therefore included in this paper (Rajan and Zingales, 1995; Booth et al., 2001; De Jong et al., 2008). The selection of the variables is motivated by agency cost theory, pecking order theory and the trade-off theory. Literature is not consistent in the selection of these variables. The decisive factor in this study is the comparability with previous research. Since all the countries that are included in this study are also analyzed in the paper by De Jong et al. (2008) this is used as comparison.
Asset tangibility indicates the amount of collateral a firm can use to secure its debt. It is measured as the net fixed assets over the book value of the total assets. Agency theory suggests that firms that are more levered tend to underinvest in order to transfer wealth away from creditors to equity holders. Lenders therefore require more collateral to secure their loans. Static trade-off theory says that more tangible firms use more debt because of reduced financial distress costs (Booth et al., 2001). Both theories imply that asset tangibility is positively related with leverage.
The static trade-off theory presumes that larger firms have a lower risk of bankruptcy and hence a lower cost of capital. Larger firms are less likely to face financial distress which would make debt more attractive (Rajan and Zingales, 1995). Firm size can be defined as the natural logarithm of total assets. This definition is often used as an inverse proxy for the probability of bankruptcy. Another interpretation is from De Jong et al. (2008) who assume that larger firms are more diversified. The combination of size and diversification results in more stable cash flows. Therefore larger firms can afford higher levels of debt. Either way, firm size is expected to be positively related to leverage.
6
growth opportunities are likely to take on more risky projects which restrict them in securing favorable debt. Moreover, firms with high growth opportunities prefer not to use debt to minimize the wealth transfer to creditors. Growth opportunities and leverage are thus expected to be negatively related.
Profitability, defined as the ratio of earnings before interest, taxes and depreciation and amortization (EBITDA) to total assets, is theorized both positively and negatively with leverage. The trade-off theory expects that firms with higher profits have lower chance of bankruptcy and thus more incentive to take on debt for a higher tax shield. However, following the pecking order theory, firms prefer to raise capital first from retained earnings, second from debt and last from issuing new equity (Myers and Majluf, 1984). The preference for internal financing is a result of the information asymmetry between owners and outside investors. More profitable firms have a larger amount of retained earnings which can be used to finance new projects. Capital structure theory mostly reports a negative relation (Rajan and Zingales, 1995; De Jong et al., 2008). Profitability is therefore expected to have a negative relationship with leverage.
Liquidity is defined as the total current assets divided by the total current liabilities, also known as the current ratio. The pecking-order theory argues that firms with high liquidity borrow less than firms with low liquidity. Besides, liquid assets can be manipulated by managers to favor shareholders. However, this is against the interest of bondholders hence increasing the agency costs of debt. The effect of liquidity on leverage is not without debate and is argued by some researchers to have a positive effect on leverage (Williamson, 1988; Sibilkov, 2009). Sibilkov (2009) has found for US firms that asset liquidity increases optimal leverage and argues that managers control for the costs of illiquidity and inefficient liquidation by adjusting leverage and the probability of incurring liquidation costs. Furthermore, the relation between asset liquidity and leverage is found to be stronger for companies that have fewer fixed assets to leverage and for companies that have a greater probability of default. Since I use the same measurement as De Jong et al. (2008), I expect liquidity to be negatively related to leverage.
7
The corporate tax rate is the last firm-specific determinant that might influence capital structure. Corporate tax rate is defined as the average tax rate of the year and is calculated as the total income taxes divided by pre-tax income. This measure is chosen instead of the marginal tax rate since this measure concerns the level of debt whereas the marginal tax rates explain incremental change in debt rather than the debt level itself (De Jong et al., 2008). The trade-off theory asserts that a major argument for using debt instead of equity is to save on corporate tax. Hence, the corporate tax rate is expected to be negatively related to leverage.
2.3. Country-specific determinants of capital structure
Determinants of capital structure can also differ due to country-related aspects. Literature on this subject not only includes a range of different variables, but often uses different terminology as well. However, there is some general overlap which makes a broader categorization possible. Namely, country-specific variables can be related to the institutional (legal), financial or macro-economic environment.
2.3.1. Institutional (legal) environment
A country’s legal, regulatory and institutional environment affects the relationship between the firm’s stakeholders and consequently the process of corporate governance (Demirgüç-Kunt and Maksimovic, 1996; La Porta et al., 1998). Classified by La Porta et al. (1998) Greece and Italy are both bank-oriented countries whereas The Netherlands and Sweden are market-oriented. The financial orientation of a country determines the importance of the banking sector and stock market. However, the impact of financial orientation on leverage is often complex. Rajan and Zingales (1995) point out that the difference between bank-oriented countries and market-oriented countries is mainly seen in the choice between public (stocks and bonds) and private financing (bank loans) than in the amount of leverage. Greater availability of debt does not necessarily lead to more borrowing since firms may not want to borrow more beyond a certain point.
8
worldwide governance indicators for more than 200 countries over almost two decades3. The governance indicators consist of six aggregate dimensions of governance of which four are found relevant for this study4.
The first institutional variable that is included concerns the effectiveness of the government, which is defined as the perception of the quality of the legal framework to formulate and implement sound policies. It is used as a substitute for the judicial system variable used in De Jong et al. (2008). A related indicator, regulatory quality, captures the perception of the government’s ability to implement policies which stimulate the private sector. The indicator is a good substitute for the legality variable used by De Jong et al. (2008). This variable measures to what degree borrowers and lenders are protected by collateral and bankruptcy laws. When creditors are better protected by the law it is easier for firms to obtain loans when they invest in intangible assets. Moreover, from the perspective of agency theory, creditors rely on the threat of not renewing short-term loans to control firms’ opportunistic behavior. Better creditor protection gives firms more access to long-term debt that would otherwise be confined to the use of short-term loans (Giannetti, 2003). Thus, both legal rights and government efficiency is hypothesized to be positively related to leverage.
Two other institutional variables are included that are related to the extent in which individuals, companies and the state respect the institutions that govern them. The first is rule of law and is an assessment of the law and order tradition in the country. It functions as a direct substitute of the rule of law variable used in De Jong et al. (2008). Corruption is another institutional aspect that provides additional information on the rule of law in a country. It is an aggregate indicator that ranks countries in terms of how corruption is perceived to exist among public officials and politicians. Outside investors that are better protected by the law become more willing to invest in firms. Firms operating in countries without such protection tend to rely more on internally generated funds. The rule of law is positively related to leverage whereas corruption is expected to be negatively related.
3 The methodology behind the calculation of these governance indicators is explained by Kaufmann et al.
(2011).
4
9
2.3.2. Financial environment
Other important variables that should be considered when analyzing the financing choices of firms concern the financial environment of a country.
Bond market development is defined as the private bond market capitalization to GDP. Faulkender and Petersen (2006) find that firms with better access to public bond markets are significantly more levered. Likewise, Doukas et al. (2011) studied the effect of a favorable debt market on capital structure choice. They find that adverse selection costs of equity at the firm level have significant impact on the choice of capital structure. When equity is out of favor firms tend to engage more in debt-financing. This engagement intensifies when debt markets are more favorable. Booth et al. (2001) also found that highly developed debt markets are associated with higher private sector debt ratios. Hence, it is expected that bond market development has a positive effect on leverage.
Stock market development is defined as the stock market capitalization over the country’s GDP. It is argued by Demirgüç-Kunt and Maksimovic (1996) that when equity markets become more developed, equity increasingly becomes a more viable option for firms to finance their investments. Debt is substituted by equity which lowers the leverage ratio of the firm. Therefore, a negative relation with leverage is expected
2.3.3. Macro-economic environment
As a measurement of general economic conditions the real GDP growth rate is included. Real economic growth positively influences the leverage of firms (Booth et al., 2001; De Jong et al., 2008). They point out that in periods with real economic growth prospect firms tend to borrow more.
Additional to the general economic conditions, capital formation is the level of gross domestic capital mobilization which can have an impact on firms’ financial decisions (De Jong et al., 2008). It consists of outlays on additions to the fixed assets of the economy plus net changes in the level of inventories. With more available funds, firms face less dependence on debt usage. Capital formation is expected to have a negative effect on leverage.
2.4. Capital structure in periods of crisis
10
subject focused on the Asian crisis of 1997 (Deesomsak et al., 2004; Fattouh et al., 2005) and found evidence that the determinants of capital structure changed for these Asian countries. Since these studies concern Asian firms which operate under different conditions these results cannot easily be generalized to European firms. Fortunately, the first studies on the financial crisis and European sovereign debt crisis have spawned recently (Alves and Francisco, 2013; Harrison and Widjaja, 2013; Iqbal and Kume, 2014, 2013). Alves and Francisco (2013) analyzed the capital structure of firms across 43 countries worldwide over three crisis periods5. Though comprehensive from a statistical point-of-view, the study by Alves and Francisco (2013) provides little theoretical explanation. Moreover, the change in capital structure determinants before and after the crisis is not explicitly discussed, only the impact crisis years have on the overall model. Harrison and Widjaja (2013) investigate the capital structure of US listed firms over the years 2004-2011. They find that the effect of tangibility is larger after the crisis. Further, they find that after the crisis the pecking order theory becomes more relevant explaining their model. However, the increased favor for pecking order theory after a crisis is contradictory to findings by Fan and So (2004)6. Iqbal and Kume (2013) do a similar study but instead focus on UK, French and German firms. Pertaining to the pre-crisis period, they find that the leverage of these firms increased in the period 2008 and 2009 but decreases afterwards. Some other studies also find that in times of financial crisis firms relied more on the use of public debt (Fosberg, 2012; Kahle and Stulz, 2013). Fosberg (2012) reported a significant increase in the debt ratios of US firms over the crisis period and a gradual decline afterwards. Overall, it is not clear how a financial crisis affects capital structure decisions. This study aims to shed more lied on the matter.
3. Data and Methodology
In this chapter first the construction of the sample is discussed after which the dependent and independent variables are defined. Furthermore, the descriptive statistics on these variables are given. This chapter concludes with the specification of the models.
5
Beside the subprime crisis and sovereign debt crisis the study by Alves and Francisco (2013) also includes the Dot.com crisis of 2000.
6 Fan and So (2004) conducted survey’s on managers of Hong Kong firms before and after the 1997 Asian crisis.
11
3.1. Sample
The selection of sample countries, Greece, Italy, The Netherlands and Sweden was discussed in the introduction. Briefly summarized this selection, (1) allows for an analysis of the impact that the financial crisis and subsequent sovereign debt-crisis had on firms’ capital structure decisions operating in each country and (2) allows for a more thorough understanding of the differences between firms’ capital structure choices in advanced but diverse European nations. The data that are used in this paper come from multiple sources. Annual firm-level data for the period 2004-2011 is retrieved from the Orbis database7. Only public companies that are listed on the relevant national stock exchange and have a continuous dataset over the period are included. Missing values are corrected for by taken the average of the previous and following year of that value. It is customary in capital structure literature to exclude financial firms from the dataset since they are under regulations that heavily influence their capital structure decisions and hence make results difficult to interpret. In the same way subsidiaries are excluded. The final sample consists of 118 Greek, 109 Italian, 60 Dutch and 125 Swedish companies.
Country-level data is retrieved from three databases that are all provided by The World Bank. Data concerning the institutional environment is available in the ‘Worldwide Governance Indicators’ database. Data on the financial environment variables is available in the ‘Global Financial Development’ database. Lastly, data on the macro-economic environment is available in the ‘World Development Indicators’ database. Using these databases provides two huge advantages:
-Consistency, which reduces the chance of measurement errors over time or between countries -Availability, data on many countries over long time-periods.
3.2. Dependent variables
Chapter 2 showed that several definitions of leverage can be used. The most commonly used leverage ratio in recent literature about capital structure is the market leverage of a firm. Book leverage ratios don’t account for appreciated and depreciated assets and thus may not give an accurate leverage ratio. Besides, studies that do check their results with book ratios consistently find the same results when using market ratios (Booth et al., 2001; De Jong et al., 2008). Therefore I only include the long-term market leverage ratio in the regression models. Book leverage is only analyzed in the univariate results section. As mentioned before, including
7
12
term debt (i.e. loans) makes results harder to interpret since short-term debt has other determinants than long-term debt (De Jong et al., 2008).
3.3. Independent variables
In chapter 2 the selection of the independent variables is already discussed. A brief overview is given in table 3.1.
Table 3.1 – Independent variables
Variable Definition
Firm-specific
Tangibility (TANG) Net fixed assets over the book value of total assets. Firm size (SIZE) Natural logarithm of total assets.
Growth (GROWTH) Market value of total assets over book value of total assets. Profitability (PROF) EBITDA over total assets.
Liquidity (LIQUID) Current assets over current liabilities.
Firm Risk (RISK) Standard deviation of operating income of previous three years. Tax (TAX) Total income tax divided by pre-tax income.
Country-specific
Enforcement (ENFORCEMENT)
The average of multiple governance indicators; Legal efficiency, Legal rights, Rule of law and Corruption. (Source: Worldwide Governance Indicators dataset)
Stock market capitalization (STOCK)
Total value of listed shares as percentage of GDP. (Source: Global Financial Development dataset)
Bond market capitalization (BOND)
Total value of private domestic debt securities issues as a percentage of GDP. (Source: Global Financial Development dataset)
Capital formation (CAPITAL) Annual gross capital formation as percentage of GDP (Source: World Development Indicators dataset)
GDP growth (GDP) Annual GDP growth rate (Source: World Development Indicators dataset)
Note: This table defines all independent variables that are used in the models. The variables’ operationalization is given in the parentheses. The data needed to calculate the firm-specific variables are all retrieved from Orbis.
3.4. Descriptive statistics
The descriptive statistics of the leverage and firm-level variables are presented in table 3.2. The numbers that are reported are averages of the total sample period. Greece has the highest mean short-term leverage while Italy had the highest mean long-term leverage. The Netherlands has the highest long-term leverage in 2005 but the lowest in 2011. The short-term leverage of The Netherlands is very low as well. Compared with the data De Jong et al. (2008)8 reported, the mean leverage ratio for each country is higher. Greek firms have a long-term leverage ratio of 0.14 now against 0.06 then. Leverage of Italian firms almost doubled from 0.08 to 0.15. Leverage of The Netherlands and Sweden only increased slightly from 0.09 to 0.12 and 0.10 to 0.13 respectively. Furthermore, changes of long-term and short-term leverage over time are depicted by figure 3.1 and figure 3.2 respectively. An interesting observation is that in 2007, just before the start of the crisis, both short-term and long-term leverage increases for all countries.
8
13
Leverage even further increases at the time of the financial crisis in 2008. After the crisis long-term leverage slowly decreases but remains above pre-crisis levels. This observation gives motivation to further investigate the impact that the financial crisis had on the determinants of capital structure.
Descriptive statistics of the country-level variables can be found in table 3.3. As depicted in the table, compared to The Netherlands and Sweden, enforcement is particularly weak in Italy and Greece9. After the crisis this even worsens for Italy and Greece. In all four countries stock markets plummet at the start of the crisis. Dutch and Swedish stock markets make a small recovery in 2010 whereas Greece’s and Italy’s stock market do not. The public bond market capitalization on the other hand gradually increased for all countries. Since the crisis, Greece’s bond market capitalization has almost quadrupled from 6.2% pre-crisis to 23.7% post-crisis whereas Italy’s bond market capitalization grew from 28.4% pre-crisis to 38.2% post-crisis10. The Dutch and Swedish bond market showed less explosive growth. It should be noted that both country’s bond market as a percentage of GDP are larger than Greece’s and Italy’s. The increase in public bond market capitalization is reflected by the increased long-term leverage ratios of firms in all countries. More large differences between the countries can be observed. Compared with the pre-crisis period GDP growth in the post-crisis period was lower for all countries. Greece’s economy was in a deep recession after the crisis with an average growth of -5.1%. Italy and The Netherlands also had a negative growth. Sweden on the other hand had an average growth of 1.5% in the post-crisis period.
9 Enforcement is the average of the standardized values of rule of law, legal efficiency, legal rights and
corruption. Over the sample period the value of all four variables decreases for Greece and Italy but stay roughly the same for The Netherlands and Sweden.
10
14 Figure 3.1 - Long-term leverage ratio.
Note: This graph depicts the development of the average long-term leverage ratio of firms categorized per country over the time period 2004-2012. Source: Own calculations based on data from Orbis.
Figure 3.2 - Short-term leverage ratio.
Note: This graph depicts the development of the average short-term leverage ratio of firms categorized per country over the time period 2004-2012. Source: Own calculations based on data from Orbis.
In table 3.4 a reduced correlation matrix is presented11. There is no strong correlation between the dependent variables and the independent variables12. The firm-specific independent
11 The complete correlation matrix including omitted variables can be found in the appendix A3.
12 It should be noted that when short-term leverage is used instead of long-term leverage the correlations
between the variables tend to increase. Since short-term leverage is not used in any of the regression models this is just a stray observation.
0 0,05 0,1 0,15 0,2 0,25 2004 2005 2006 2007 2008 2009 2010 2011 2012
Long-term leverage ratio
GR (LTD) IT (LTD) NL (LTD) SE (LTD) 0 0,05 0,1 0,15 0,2 0,25 0,3 2004 2005 2006 2007 2008 2009 2010 2011 2012
Short-term leverage ratio
15
variables all have a low correlation with each other. The country-specific variables on the other hand are more correlated. Corruption, legal rights, rule of law and efficiency are all highly correlated. There is also a moderate correlation between the stock market capitalization and the bond market capitalization. Also most of the country-specific variables are moderately correlated with GDP growth. All in all, high correlation between these variables could lead to severe problems in the regression analysis. In order to prevent potential multicollinearity in the model, a new law variable is created. This procedure is similar as used in other studies that have this problem (e.g. De Jong et al., 2008). The new overarching law-variable named law enforcement (ENFORCEMENT) is calculated as the average of the standardized values of legal rights, legal efficiency, rule of law and corruption. After the creation of this variable there is still a high correlation with both stock market capitalization and bond market capitalization and this new variable. This should be accounted for when interpreting the results.
Table 3.2 - Firm-level descriptive statistics
LTDLEV TANG SIZE GROWTH PROFIT LIQUID RISK TAX
All countries Mean 0.1389 0.2633 5.2921 1.4332 0.0829 1.8913 0.0611 0.3209 Median 0.0992 0.2054 5.2307 1.1139 0.0844 1.3845 0.0288 0.2555 Maximum 0.9046 0.9987 8.0369 22.8068 1.1894 87.8960 6.4555 185.8000 Minimum 0.0000 0.0000 0.6990 0.1252 -10.7149 0.0110 0.0000 -77.0000 Std. Dev. 0.1444 0.2313 0.9450 1.2186 0.2472 3.7453 0.2315 4.2650 Greece Mean 0.1485 0.3795 4.8326 1.1928 0.0645 2.2938 0.0435 0.5869 Median 0.1088 0.3564 4.7936 0.9497 0.0608 1.4760 0.0247 0.2433 Maximum 0.7041 0.9391 7.0165 11.1096 0.8668 87.8960 1.5696 185.8000 Minimum 0.0000 0.0000 0.6990 0.2976 -0.7573 0.0880 0.0001 -12.2771 Std. Dev. 0.1495 0.2281 0.8248 1.0033 0.1081 6.1532 0.0940 7.1167 Italy Mean 0.1514 0.2703 5.5976 1.2563 0.0863 1.4301 0.0474 0.3353 Median 0.1195 0.2122 5.5023 1.1098 0.0861 1.2460 0.0204 0.3349 Maximum 0.8360 0.9459 8.0369 7.7540 1.1894 7.9420 2.8398 30.5521 Minimum 0.0000 0.0000 3.4012 0.1252 -4.6222 0.1480 0.0000 -39.2813 Std. Dev. 0.1358 0.2182 0.8079 0.6369 0.1900 0.8388 0.1572 2.3042 The Netherlands Mean 0.1175 0.2010 5.7163 1.5001 0.0962 1.5004 0.1000 0.1738 Median 0.0945 0.1555 5.7590 1.2604 0.1188 1.3420 0.0325 0.2241 Maximum 0.5987 0.7644 7.6947 9.4860 1.0602 17.9290 6.4555 1.9359 Minimum 0.0000 0.0000 1.8451 0.4432 -10.7149 0.0430 0.0000 -4.6923 Std. Dev. 0.1194 0.1788 0.9415 0.9688 0.5389 1.1911 0.5390 0.4231 Sweden Mean 0.1292 0.1775 5.2560 1.7821 0.0911 2.1012 0.0709 0.1280 Median 0.0728 0.0899 5.1963 1.2747 0.1000 1.4740 0.0442 0.2468 Maximum 0.9046 0.9987 7.5419 22.8068 0.5582 50.9070 0.5217 11.6121 Minimum 0.0000 0.0000 1.9986 0.4875 -0.9465 0.0110 0.0005 -77.0000 Std. Dev. 0.1557 0.2203 0.9699 1.7144 0.1390 2.9634 0.0797 2.7140
16 Table 3.3 - Country-level descriptive statistics
Enforcement Stock Bond Capital %GDP
Greece pre-crisis 0,67 67,96 6,16 24,26 3,77 post-crisis 0,42 20,63 23,68 17,41 -5,06 Italy pre-crisis 0,54 48,41 28,38 21,60 1,60 post-crisis 0,43 17,31 38,19 19,56 -1,11 The Netherlands pre-crisis 1,85 102,58 68,78 19,82 3,12 post-crisis 1,87 70,65 74,32 18,10 -0,40 Sweden pre-crisis 1,85 120,26 43,92 18,93 3,59 post-crisis 2,00 98,39 57,81 18,36 1,49
Note: This table shows the how the country-specific variables have changes after the start of the crisis. Definitions of the variables can be found in table 3.1. The pre-crisis and post-crisis values are averages over the period 2005-2007 and 2009-2011 respectively. A more extensive table that includes the development of each of these values over time can be found in the appendix A1.
Table 3.4 - Correlation matrix
Tang Size Growth Profit Liquid Risk Enforce Stock Bond Capital %GDP
Dependent variable Long-term leverage 0.222 0.189 -0.260 -0.048 -0.103 -0.046 -0.089 -0.130 -0.037 -0.024 -0.129 Independent variables Tang 1 Size 0.002 1 Growth -0.130 -0.075 1 Profit -0.005 0.169 0.079 1 Liquid -0.106 -0.175 0.035 -0.031 1 Risk -0.095 -0.081 0.087 -0.519 0.019 1 Enforcement -0.288 0.078 0.190 0.036 0.011 0.071 1 Stock -0.237 0.032 0.241 0.047 0.020 0.062 0.858 1 Bond -0.301 0.237 0.100 0.029 -0.040 0.080 0.755 0.466 1 Capital 0.131 -0.035 -0.026 0.025 -0.013 -0.026 -0.339 0.031 -0.551 1 %GDP -0.117 0.065 0.177 0.058 -0.006 0.022 0.337 0.640 0.058 0.484 1 3.5. Model specification
To assess the determinants of capital structure in the sample countries, firms’ leverage ratios are modeled as a function of several firm-specific factors. For each country independently and for the total sample the relationship is estimated using the following empirical model:
𝐿𝐸𝑉i,t = αi,t + 𝛽1TANGi,t + 𝛽2SIZEi,t + 𝛽3GROWTHi,t + 𝛽4𝑃𝑅𝑂𝐹ITi,t + 𝛽5LIQUIDi,t + 𝛽6RISKi,t +
𝛽7TAXi,t +𝜀i,t (1)
17
error term for ith firm at time t and αi,t is the regression intercept which can vary across firms
and over time.
To test for country-specific effects a common practice is to add country-dummies to the model (Booth et al., 2001; Deesomsak et al., 2004). However, this method is incapable of revealing what parts of the institutional environment were affected by the crisis. To explain where differences in the capital structure decision between countries come from and to understand what drives these differences the underlying country-specific variables should be analyzed. Mainly following De Jong et al. (2008) I add a selection of country-specific variables to see if the explanatory power of the model increases and what the effect of the crisis was on these variables. This results in the following model:
𝐿𝐸𝑉i,t = αi,t + 𝛽1TANGi,t + 𝛽2SIZEi,t + 𝛽3GROWTHi,t + 𝛽4𝑃𝑅𝑂𝐹ITi,t + 𝛽5LIQUIDi,t + 𝛽6RISKi,t +
𝛽7TAXi,t + 𝛽8ENFi,t + 𝛽9STOCKi,t + 𝛽10BONDi,t + 𝛽11CAPITALi,t + 𝛽12ΔGDPi,t + 𝜀i,t (2)
It is unlikely that the empirical capital structure models discussed above are fully specified (Booth et al., 2001). To account for the problem of omitted explanatory variables fixed effects panel data can be used. Specifically I use fixed cross-sectional effects in both the firm-specific and country-specific model.
In order to reveal the possible impact the financial crisis had, the fixed-effects panel data analysis, for both the firm and country models, is done over two different periods; pre-crisis and post-crisis (Deesomsak et al., 2004)13. The pre-crisis sample includes data on the years 2005-2007. The year 2004 is excluded since many firms had missing values in that year which would have resulted in an unbalanced panel. The post-crisis sample includes the years 2009-2011. The year 2012 is excluded since not all data on the country-level variables is available which makes a comparison between the firm and country model more difficult. Besides, both panel models now have the same amount of observations. Since a crisis year can have an unwanted impact on the results the year 2008 is not included. Comparison of the pre-crisis and post-crisis model is done by interpreting the significance of the variables, discussing the explanatory power of the model and testing the change in size of the coefficients.
13
18
4. Results
The results are reported in three parts. The first section presents univariate results concerning the leverage ratios. In the second section the firm-specific results are discussed. In the final section of this chapter the impact of adding the country-specific variables to the model is analyzed.
4.1. Univariate results
The previous chapter concluded that firms’ long-term leverage ratio increased after the crisis. To test whether the samples of leverage and the independent variables indeed significantly differ from pre-crisis to post-crisis, paired t-tests are conducted. The results are presented in table 4.1. The value of the pre-crisis and post-crisis period is calculated as the average of the years 2005-2007 and 2009-2011 respectively. Beside leverage, t-tests are also conducted on the input of the ratios; market value of equity and long-term debt. The null hypothesis in all cases is that the difference between the samples is zero (i.e. there is no significant evidence that one sample is different from the other sample). The alternative hypothesis being that the difference is larger than zero. These t-tests are relevant since this study examines the impact the crisis had on capital structure and its determinants.
From table 4.1 we can infer that the long-term leverage ratio of all countries is significantly larger in the post-crisis period with respect to the pre-crisis period. Since I use market values for equity it is possible that this change in leverage ratio is a result of declining market values due to plummeting stock markets. More paired t-tests show that the book leverage of all countries is higher in the post-crisis than in the pre-crisis. However, these results are only found to be significant in Greece and Italy14. Additionally, also significant values for short-term leverage ratios are found. Furthermore table 4.1 shows that firms’ long-term debt has significantly increased in all countries after the crisis. The market value of equity has significantly decreased in all countries but Sweden. Book value of equity on the other hand increased in all countries. Furthermore, the increase in Dutch and Swedish firms’ their leverage can be attributed mainly to the drop in their market value whereas Greek and Italian firms indeed took on a higher leverage ratio.
The t-tests are also performed for the explanatory variables of the total sample. The complete results can be found in the appendix B1. For long-term leverage of the total sample the
14
19
crisis and pre-crisis sample are again significantly different from each other. The t-tests performed on firm-specific variables on the other hand could not reject the null hypothesis that there is no difference between the samples. The only exception being growth opportunities, for which a significant difference was found.
Table 4.1 – Paired t-tests
Greece Italy Netherlands Sweden
pre post pre post pre post pre post
Long-term leverage mean 0.1139 0.1738 0.1237 0.1737 0.0970 0.1222 0.1106 0.1305 t-stat 5.6182*** 5.7581*** 2.8676*** 2.9040*** LTD mean 107312 140828 1053034 1567316 539242 843729 314806 486636 t-stat 2.2518** 1.7174** 2.5345*** 3.1885*** Equity mean 459512 182508 2702758 1928544 2928750 2196694 1806467 1811677 t-stat 3.1614*** 2.1483** 2.2934** 0.0364
Note: This table presents the means and t-tests results for differences in mean of different variables spread out per country across two periods, pre-crisis and post-crisis. Long-term leverage is the market long-term leverage ratio, LTD is the total amount of long term debt and Equity represents the market value of equity. The ***, ** and * represent a statistical significance at 1%, 5% and 10% levels respectively.
4.2. Firm-specific results
The results of the firm-specific model for the full sample are shown in table 4.2. The model is estimated for the pre-crisis and post-crisis period. Based on the theory discussed in the chapter 2, in the last column the predicted sign of the coefficients is given. First the model that includes all countries is presented after which a more in-depth analysis of each country is given.
4.2.1. Panel results for all countries sample
Tangibility. In line with capital structure theory, a significant effect for tangibility as well as a
large positive coefficient is found in the pre-crisis model. In the post-crisis model the positive relation of tangibility and leverage becomes larger and is even highly significant. A Wald test shows that the coefficient of the post-crisis model is indeed significantly different from the pre-crisis model from which we can infer that after the pre-crisis tangibility has become a more important determinant of leverage.
Firm size. In the pre-crisis model there is also a highly significant positive effect of firm size on
20 Growth opportunity. A highly significant negative effect for growth opportunity on leverage has
been found in the pre-crisis model as well. The coefficient is smaller than the coefficients of tangibility and firm size but still has some explanatory power. After the crisis the effect of growth opportunity on leverage diminishes. A Wald test shows that the post-crisis coefficient is indeed significantly different from the pre-crisis coefficient.
Profitability. As expected, profitability is negatively related with leverage. However, the effect is
small and insignificant.
Liquidity. A small, but highly significant, positive effect is found for the liquidity of a firm. The
positive effect becomes larger in the post-crisis model.
Risk. Before last, this model has found no significant evidence that risk is related to leverage. Tax rate. Finally, also almost no effect of tax rate on leverage is found.
Overall, for the model including all countries, the signs of the coefficients before and after the crisis are largely in line with contemporary literature on capital structure. However, the results also show that the crisis indeed altered the relation that of some of the variables have with leverage. As we see now this differs per country.
4.2.2. Panel results for individual countries
The fixed-effects panel analysis with firm-specific determinants is also done independently for each country. The signs of most coefficients confirm corporate theories discussed in chapter 2. However, there are differences between countries. For a start, the significance of variables between countries differs widely, which encourages investigating the role of the country-specific variables which is done in the next section. Moreover, the role of the firm-specific variables is subject to change when the post-crisis model is compared to the pre-crisis model.
21
and The Netherlands. In Greece the effect was very strongly negative related which is conform the theory that riskier firms have lower leverage. The Netherlands on the other hand shows a positive effect. A possible explanation could be that the measurement of firm risk in this paper is not correct. Problems with this variable also arise in other studies (Deesomsak et al., 2004; De Jong et al., 2008). The same goes for tax for which the results are negligible. Overall the explanatory power15 of all models increases after the crisis with Greece being the exception. In the Greek post-crisis model many of the coefficients become smaller, flip in sign or lose significance. Wald statistics indeed find significantly different values for four of the coefficients hinting that the crisis had a severe impact on the Greek capital structure model.
4.3. Country-specific results
The results of the fixed-effects panel data models with country-specific variables are shown in table 4.316. It should be noted that these results only say something about capital structure determinants of the sample countries as a whole. First it is checked if adding the country-specific variables to the model increases the explanatory power and if these variables are able to predict capital structure17. Second, the impact of the financial crisis on the country-specific determinants is assessed.
The adjusted R-squared indicates that the explanatory power of the model does increase but only very slightly. Overall, the sign and significance of the firm-specific variables discussed in the previous section stay the same in the new model. The country-specific variables thus are able to predict some of the variance of leverage but less than expected. Two of the possible explanations are now given. First, the number of countries that are included in the model might be too low which results in lack of variation in the variables. Similarly the number of observed time periods might be too low. The model is rerun over the total time period but this failed to enhance the results. However, this could also be attributed to the fact that the two periods are different from each other. Booth et al. (2001) recognize the aforementioned problem of too little countries or time-periods in country-specific capital structure models. Secondly, the
15
The explanatory power is measured by the adjusted R-squared.
16
The variable tax is excluded from this model. There are two reasons to do so. First, the variable had a contradicting effect in the firm-specific models and overall a very small coefficient. Moreover the variable was never significant. Second, due to differences in tax regulations between countries interpretation of the results is more difficult. Dropping the variable from the country-specific models has been tested and the results are indeed stronger without the variable.
17
22
measurement that is used for the variables may be incorrect. Though largely in line with De Jong et al. (2008), who do find significant results for these variables, the variables might not capture the institutional difference between the countries correctly. Since De Jong et al. (2008) use a different source for the institutional variables this is plausible.
With the above-mentioned limitations in mind some observations regarding the impact of the crisis are now discussed. From table 4.3 can be inferred that the crisis seems to have impacted the country-specific determinants of capital structure in different ways. First, the level of enforcement has a, theoretically incorrect, negative effect on leverage in the pre-crisis model. This effect even becomes larger in the post-crisis model. Also, pre-crisis, stock market and bond market have respectively a significant small negative and significant small positive effect on leverage. This is what we would expect. In the post-crisis the effect becomes negative for bond market but loses significance. Capital mobilization and GDP growth both have insignificant and negligible effects on leverage.
Table 4.2 - Panel analysis of the firm-specific determinants of leverage
Variables All countries Greece Italy Netherlands Sweden Predicted
pre-crisis post-crisis pre-crisis post-crisis pre-crisis post-crisis pre-crisis post-crisis pre-crisis post-crisis Value
Constant -0.1255* -0.0496 -0.4536*** -0.0471 -0.2444 -0.5794** 0.1492 0.0952 -0.0377 -0.2027 (t-statistic) (-1.8862) (-0.6235) (-3.0380) (-0.3031) (-1.3391) (-2.4652) (0.7542) (0.7441) (-0.4505) (-1.4601) Wald test 0.9543 2.6135*** 1.4252 0.4218 1.1883 TANG 0.0856** 0.1948*** 0.0576 0.1875 0.1592** 0.2562*** -0.0516 0.2027* 0.2144** 0.1937* + (t-statistic) (1.9779) (4.0004) (0.7177) (1.0476) (1.9734) (4.9311) (-0.5346) (1.7487) (2.3219) (1.6881) Wald test 2.2422** 0.7260 1.8669* 2.1936** 0.1803 SIZE 0.0439*** 0.0240* 0.1123*** 0.0161 0.0623* 0.1152*** 0.0023 0.0025 0.0282* 0.0577** + (t-statistic) (3.6416) (-1.6592) (3.8880) (0.5893) (1.9619) (2.8161) (0.0658) (0.1126) (1.8761) (2.2681) Wald test 1.3788 3.5311*** 5.7915*** 0.0086 1.1608 GROWTH -0.0151*** -0.0050 -0.0140 -0.0113 -0.0417*** -0.0644*** -0.0308*** -0.0047 -0.0068* -0.0050 - (t-statistic) (-4.5883) (-1.2792) (-1.1666) (-0.3960) (-4.1570) (-2.9438) (-5.0764) (-0.4918) (-1.8738) (-1.5851) Wald test 2.6088*** 0.0951 1.0371 2.7335*** 0.5774 PROF -0.0068 -0.0163 0.0234 0.0668 -0.0442** 0.0690 0.0019 -0.1461** -0.0849** -0.0922** - (t-statistic) (-1.0565) (-0.3952) (0.3117) (0.5804) (-2.2721) (0.4584) (0.3669) (-1.9940) (-2.2833) (-2.0273) Wald test 0.2290 0.3771 0.7517 2.0196** 0.1609 LIQUID 0.0047** 0.0167*** 0.0155*** 0.0317*** 0.0261*** 0.0781*** -0.0012 -0.0019 -0.0049 0.0025 - (t-statistic) (2.1589) (6.2088) (3.8093) (5.1043) (3.4832) (7.0941) (-0.3444) (-0.2719) (-1.4496) (1.0529) Wald test 4.4555*** 2.6067*** 4.7274 0.1015 3.0922*** RISK 0.0172 0.0154 -0.5030*** 0.0126 0.0161 0.0426 0.0268*** -0.0414 -0.0726 0.0471 - (t-statistic) (1.4622) (0.3439) (-3.0708) (0.1334) (0.2420) (0.4398) (3.0797) (-0.5362) (-1.0879) (0.8037) Wald test 0.1928 86.0265*** 0.2736 0.8829 2.0422** TAX -0.0002 0.0003 0.0003 0.0003 0.0014 0.0055* -0.0015 0.0034 -0.0009 0.0015 + (t-statistic) (-0.2951) (0.8216) (0.1474) (0.5344) (1.2322) (1.7268) (-0.2046) (0.3207) (-1.2269) (0.3003) Wald test 1.2711 0.0387 1.3001 0.4611 0.4822 Adj R² 0.7960 0.8103 0.7663 0.6703 0.7177 0.8439 0.8595 0.9126 0.8866 0.9266 Obs 1236 1236 354 354 327 327 180 180 375 375
Table 4.3 - Firm-specific and country-specific determinants of leverage
Note: This table presents regression results of leverage (LEV) on firm-specific and country-specific variables for all sample countries using equation 2. The regressions are done across two periods, pre-crisis and post-crisis. All variables are defined in chapter 3. In the last column the predicted sign of the coefficient is given. The t-statistics are reported in parentheses. The ***, ** and * represent a statistical significance at 1%, 5% and 10% levels respectively.
5. Discussion
This chapter compares and discusses whether the results of the previous chapter are in line with previous research. Similar as Deesomsak et al. (2004), Alves and Francisco (2013) and Iqbal and Kume (2013), I have found that firms’ long-term leverage increases after the crisis. In the period after the crisis firms were significantly more levered than before the crisis. These findings are
Variables All countries predicted
pre-crisis post-crisis value
25
visible in all sample countries included in this study. A part of the increase in long-term leverage can be attributed to the fact that market equity dropped for these firms. However, for Italy and Greece, two bank-oriented countries that experienced a sovereign debt crisis over the initial financial crisis, also a significant increase in book leverage was found indicating that apart from the drop in market value these firms still became more levered. Due to the crisis the stock markets collapsed in all sample countries. This was especially the case in Italy and Greece. Whether the increase in leverage is due to an underdeveloped stock market in these countries or firms’ choice to take on more debt cannot be concluded. Kahle and Stulz (2013) have found that net equity issuance fell sharply in the first year of the crisis whereas net debt issuance did not. This could explain why Greek and Italian firms became more levered. For firms from Sweden and The Netherlands, the increase in leverage can mostly be attributed to the drop in market leverage.
Apart from the crisis the firm-specific results in general correspond to those reported by De Jong et al. (2008). There are some differences. They have found more support for the effect of risk and tax. I almost find no significant support for these variables. I also have found that for Greece and Italy liquidity has a significant positive effect on leverage whereas De Jong et al. (2008) reports negative signs. These results can best be explained by the findings of Sibilkov (2009). Greek and Italian firms who had more liquid assets post-crisis were better able to increase their leverage. After the crisis the probability of default increased for firms of these countries. It is therefore possible that, additional to fixed assets, liquid assets were increasingly used as collateral. In Sweden and The Netherlands liquidity of assets did not become more important after the crisis. Similar as Deesomsak et al. (2004) and Alves and Francisco (2013) I also find that the capital structure model has been affected by the crisis. Like Harrison and Widjaja (2013), the effect of tangibility on leverage significantly increased after the crisis. This can be explained by the fact that bank and investors demand more collateral in times of crisis. The effects of growth opportunity and firm size become less important after the crisis. Worth mentioning is that Sweden, a market-oriented country and not part of the EU, almost showed no change in its capital structure model.
26
findings of Alves and Francisco (2013) that the role of country-specific variables has altered due to the crisis. The previous chapter showed that the variable enforcement does not correctly reflect the protection of the rights of bondholders and might be badly constructed. Consequently, the effect on the model is that the better enforced countries, Sweden and The Netherlands, had better stock markets to begin with and a quicker recovery of those markets as well which has a strong effect on the model. I suspect that the multicollinearity found in chapter 3 has a part in this which makes the results difficult to interpret.
Summarized, I find that, consistent with other studies, firms’ leverage increases after the crisis. The capital structure model is also affected by the impact of the crisis. And finally, that the results of the country-specific variables are inconclusive.
6. Conclusion
This paper has shed light on how firms’ capital structure was affected by the financial crisis by comparing the leverage ratio of the pre-crisis period (2005-2007) with the post-crisis period (2009-2011). The sample used in this study consists of 412 listed firms from Greece, Italy, The Netherlands and Sweden. The average market long-term leverage ratio was found to be significantly higher in the post-crisis period for all countries. The book long-term leverage ratio was only found to be significantly larger in Greece and Italy. Furthermore, using main capital structure theories, I have analyzed how the financial crisis has impacted the firm- and country-specific determinants of capital structure. Overall, the results are in line with other capital structure studies such as Booth et al. (2001) and De Jong et al. (2008). More specific, the impact of the crisis differs per country. The effect of tangibility was found to be stronger after the crisis in Italy and The Netherlands. Also a positive effect of liquidity was found in Italy and Greece suggesting the increased importance of collateral. Finally, I have found that the country-specific determinants were also affected by the financial crisis. However the results were inconclusive and should be a concern for further research.
27
included in the sample). Furthermore it is interesting to see how firms’ their leverage will develop over a longer time period after the crisis. Does it return to the status quo ante or did the crisis permanently change firms’ capital structure. Finally, this research showed that the impact of the crisis differed substantially for each country. To get a better understanding what drives these differences similar research should be conducted for other countries.
Acknowledgments
28
References
Alves, P., & Francisco, P. (2013). The Impact of Institutional Environment in Firms’ Capital Structure during the Recent Financial Crisis.
Antoniou, A., & Paudyal, K. (2002). Determinants of Corporate Capital Structure : Evidence from European Countries Determinants of Corporate Capital Structure : Evidence from European Countries, 2, 23–26.
Booth, L., Aivazian, V., Demirguc-Kunt, A., & Maksimovic, V. (2001). Capital Structures in Developing Countries. The Journal of Finance, 56(1), 87–130.
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 & Finance, 32(9), 1954–1969.
Deesomsak, R., Paudyal, K., & Pescetto, G. (2004). The determinants of capital structure: evidence from the Asia Pacific region. Journal of Multinational Financial Management, 14(4-5), 387–405.
Demirgüç-Kunt, A., & Maksimovic, V. (1996). Stock market development and financing choices of firms. The World Bank Economic Review, 351–369.
Doukas, J. a., Guo, J. M., & Zhou, B. (2011). “Hot” Debt Markets and Capital Structure. European
Financial Management, 17(1), 46–99.
Economist, The. (2013, September 7). The origins of the financial crisis: Crash course. Retrieved from www.economist.com
Fan, D. K. K., & So, R. W. (2004). What managers think of capital structure: the evidence from Hong Kong. Journal of Asian Economics, 15(4), 817–830.
Fattouh, B., Scaramozzino, P., & Harris, L. (2005). Capital Structure in South Korea: A Quantile Regression Approach. Journal of Development Economic, 76(1), 231–250.
Faulkender, M., & Petersen, M. (2006). Does the source of capital affect capital structure?
Review of Financial Studies, 19(1), 45–79.
Fosberg, R. H. (2012). Capital structure and the financial crisis. Journal of Finance and
29
Giannetti, M. (2003). Do better institutions mitigate agency problems? Evidence from corporate finance choices. Journal of Financial and Quantitative Analysis, 38(01), 185–212.
Harrison, B., & Widjaja, T. (2013). Did the financial crisis impact on the capital structure of firms? Iqbal, A., & Kume, O. (2013). Impact of financial crisis on capital structure of UK firms.
Proceedings of the 20th Annual Conference of the Multinational Finance Society 2013.
Iqbal, A., & Kume, O. (2014). Impact of financial crisis on firms’ capital structure in UK, France, and Germany. Multinational Finance Journal, Forthcoming. Available at SSRN: http://ssrn.com/abstract=2472669.
Kahle, K. M., & Stulz, R. M. (2013). Access to capital, investment, and the financial crisis. Journal
of Financial Economics, 110(2), 280–299.
Kaufmann, D., Kraay, A., & Mastruzzi, M. (2011). The Worldwide Governance Indicators: Methodology and Analytical Issues. Hague Journal on the Rule of Law, 3(02), 220–246.
La Porta, R., López-de-Silanes, F., Shleifer, A., & Vishny, R. (1998). Law and Finance. Journal of
Political Economy, 106(6).
Lane, P. R. (2012). The European Sovereign Debt Crisis. Journal of Economic Perspectives, 26(3), 49–68.
Modigliani, F., & Miller, M. H. (1958). The cost of capital, corporation finance, and the theory of investment. The American Economic Review, 48(3), 261–297.
Myers, S. (1977). Determinants of corporate borrowing. Journal of Financial Economics, 5, 147– 175.
Myers, S. C., & Majluf, N. S. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics, 13(2), 187–221. Rajan, R., & Zingales, L. (1995). What do we know about capital structure? Some evidence from international data. The Journal of Finance, 50(5), 1421–1460.
Sibilkov, V. (2009). Asset liquidity and capital structure. Journal of Financial and Quantitative
30
Williamson, O. E. (1988). Corporate finance and corporate governance. The journal of finance, 43(3), 567-591.
31
Appendix
Table A1 - Descriptive statistics country-level variables
Efficien cy Rule of law Legal rights Corrupti on Enforce
ment Stock Bond Capital %GDP
32 Figures A2 – Development of selected country-level variables over time
0 20 40 60 80 100 120 140 160 2004 2005 2006 2007 2008 2009 2010 2011
Stock market capitalization (as % of
GDP)
GR IT NL SE 0 10 20 30 40 50 60 70 80 90 2004 2005 2006 2007 2008 2009 2010 2011Bond market capitalization (as % of
GDP)
33 Table A3 – Full correlation matrix
Tang Size Growth Profit Liquid Risk Tax Efficiency Legal Rule Corruption Stock Bond Capital %GDP
34 Table B1 - paired t-tests of the total sample
LTD TANG SIZE GROWTH PROF LIQUID RISK TAX
Sample All All All All All All All All
Observations 412 412 412 412 412 412 412 412 Sample Mean 0.1531 0.2591 5.2938 1.2624 0.0741 1.9130 0.0568 0.4160 Std. Dev. 0.1431 0.2301 0.9639 1.0143 0.0907 4.8063 0.0826 3.3169 Method t-statistic 8.7179 -1.5283 1.1649 -8.7460 -1.6312 -0.2648 -0.4569 1.1229 probability 0.0000 0.1272 0.2447 0.0000 0.1036 0.7912 0.6479 0.2620
Note: This table shows the t-tests done for the dependent variable and all the independent variables over the total sample.
Table B2 – paired t-tests of the individual samples
Greece Italy Netherlands Sweden
pre post pre post pre post pre post
Mark lev mean 0.1139 0.1738 0.1237 0.1737 0.0970 0.1222 0.1106 0.1305 t-stat 5.6182*** 5.7581*** 2.8676*** 2.9040*** Book lev mean 0.1322 0.1554 0.1544 0.1799 0.1412 0.1434 0.1472 0.1495 t-stat 2.2319** 3.0281*** 0.2334 0.3392 Short lev mean 0.1344 0.2158 0.0846 0.1331 0.0489 0.0603 0.0425 0.0550 t-stat 6.4557*** 6.0837*** 2.2240** 2.4003*** LTD mean 107312 140828 1053034 1567316 539242 843729 314806 486636 t-stat 2.2518** 1.7174** 2.5345*** 3.1885*** BookEquity mean 192477 204431 1280218 1668233 1315425 1377738 887441 1016957 t-stat 0.6023 1.8904** 0.4184 3.7353*** MarkEquity mean 459512 182508 2702758 1928544 2928750 2196694 1806467 1811677 t-stat 3.1614*** 2.1483** 2.2934** 0.0364
35 Table C1 - Panel analysis of firm-specific determinants of leverage with country dummies
Dependent Variable: LTDLEV Dependent Variable: LTDLEV
Method: Panel Least Squares Method: Panel Least Squares
Sample: 2005 2007 Sample: 2009 2011
Periods included: 3 Periods included: 3
Cross-sections included: 412 Cross-sections included: 412
Total panel (balanced) observations: 1236 Total panel (balanced) observations: 1236
Variable Coefficient
Std.
Error t-Statistic Prob. Variable Coefficient
Std.
Error t-Statistic Prob.
C 0.0365 0.0227 1.6091 0.1079 C -0.0516 0.0289 -1.7874 0.0741 TANG 0.1061 0.0154 6.8786 0.0000 TANG 0.1225 0.0191 6.4177 0.0000 SIZE 0.0210 0.0039 5.4188 0.0000 SIZE 0.0404 0.0050 8.0260 0.0000 GROWTH -0.0233 0.0026 -8.9670 0.0000 GROWTH -0.0185 0.0036 -5.2118 0.0000 PROFIT -0.0229 0.0122 -1.8789 0.0605 PROFIT -0.2243 0.0436 -5.1489 0.0000 LIQUID -0.0018 0.0012 -1.5470 0.1221 LIQUID -0.0020 0.0009 -2.3077 0.0212 RISK -0.0105 0.0151 -0.6952 0.4870 RISK 0.0150 0.0483 0.3110 0.7559 TAX -0.0002 0.0011 -0.1411 0.8878 TAX -0.0004 0.0007 -0.6370 0.5242 NETHERLANDS -0.0333 0.0105 -3.1537 0.0017 NETHERLANDS -0.0314 0.0131 -2.4017 0.0165 ITALY -0.0191 0.0089 -2.1409 0.0325 ITALY 0.0080 0.0111 0.7151 0.4747 GREECE -0.0247 0.0092 -2.6832 0.0074 GREECE 0.0187 0.0119 1.5756 0.1154 R-squared 0.1533 Mean dependent var 0.1130 R-squared 0.1520 Mean dependent var 0.1531 Adjusted
R-squared 0.1463 S.D. dependent var 0.1235
Adjusted
