Capital structure and cultural clusters
around the world
Faculty of Business and Economics, University of Groningen
Master Thesis
Author: Matthias Pieter Tamerus
Student number: s1890514
Study Program: MSc International Financial Management
Supervisor: dr. V. (Victoria) Purice
Abstract. This study aims to find to what extent culture influences capital structure around the world. In this study the effect of culture on capital structure is investigated through grouping countries into cultural clusters. The differences in the mean leverage ratio across the nine cultural clusters are being examined. Results indicate the cultural clusters have a significant impact on capital structure around the world. Furthermore, through comparing results with previous research, this study found that culture is of less importance in explaining differences in capital structure across European cultural clusters than two decades ago.
2
1
Introduction
Studies in the past have shown that managerial behaviour influences a firm’s capital structure and in turn, capital structure influences firm performance. Evidently, not all managers behave in the same way and they tend to make different choices, especially those from different cultures. This may result in managerial decisions and strategies being culturally oriented and therefore internationally different. Since a firm’s capital structure is an accumulation of managerial decisions, capital structure may be unique to cultures as well. This study will therefore build on existing literature and investigate the effect of culture on capital structure.
Over the past decades much research has been done as to what determines a firm’s capital structure. Research in this area started with Modigliani and Miller’s (1958) study, in which the authors present their findings on capital structure. They find that in a world without taxes, bankruptcy costs, agency costs and asymmetric information the value of a firm should not be affected by the firm’s capital structure. However, in today’s world these factors do exist. Many studies investigating the optimal capital structure that do take these factors into account have followed. For instance, Barton and Gordon (1988) find that managerial behaviour influences capital structure and Hirshleifer and Thakor (1992) find that agency theory also plays a role in capital structure theory. Also, Myers and Majluf (1984) state that the pecking order theory can be used to explain capital structure theory too. Besides these factors that play at a firm level, country level determinants are found to be important in determining a firm’s capital structure as well. Determinants that play at a country level are for instance legal systems, economic factors and a country’s institutional environment (De Jong, Kabir and Nguyen, 2008; Demirgüç-Kunt and Maksimovic, 1999).
3
However, a limitation of this study’s sample is that only retail firms from Europe were included and data from 1994 was used. This limits the representability of this study, since firms from other industries and other cultures were excluded. Also, there is less diversity in cultures in Europe than there is around the world. In addition to that, Bruno and Shin (2014) find that financial conditions and financial risk taking across sectors and regions are synchronizing. Since the publication of the previously mentioned study member states of the Economic and Monetary Union of the European Union (EU) implemented the euro and the EU’s influence on its members’ legislation has grown. Therefore one might propose that the capital structure of firms in Europe are less different from one another than they were over two decades ago. It is however expected that on a global scale capital structure decisions of firms around the world still differ.
With that in mind, this study seeks to find whether the effect of culture still plays a significant role in capital structure decisions in Europe and around the world. Closely related to Gleason et al. (2000), this study extents on current literature by expanding the scope of research on cultural clusters from clusters in Europe to clusters around the world. Also, Demirgüc-Kunt and Maksimovic (1999) argue that the influence of some institutional factors on capital structure is different for small firms than large firms. However, many studies only include large listed companies which may create a bias in literature (Giannetti, 2003). Moreover, Song and Philippatos (2004) argue that capital structure may be industry specific. For that reason, this study includes both large and small companies from multiple industries. Through doing so findings will be based on a representative, diverse and global sample.
In order to shed some light on international financing decisions by managers, this study’s purpose is to find to what extent culture influences capital structure by examining cultural clusters around the world. In addition, this study will investigate whether the impact of culture on capital structure decisions in the European cultural clusters still holds compared to two decades ago. This will be done by examining recent firm data from countries all over the globe. These firm’s countries will be grouped into nine cultural clusters and while controlling for other known determinants of capital structure, the differences in the mean capital structure of clusters around the world will be tested. Also, the differences between the European clusters will be compared with the differences two decades ago.
4
2
Theoretical background
This section provides a theoretical background on what determines a firm’s capital structure. Capital structure can be defined as the relevant amounts of debt and equity in a firm (Milton and Raviv, 1991) and is often referred to as debt ratio or leverage (ratio). In this study these will be used interchangeably. There are two types of factors that determine the ideal balance of a firm’s debt and equity. These can be categorized as either firm specific or country specific. In section 2.1 the firm specific determinants will be discussed and in section 2.2 the country specific determinants.
2.1 Firm specific determinants
The majority of the existing literature focuses on the firm specific determinants of capital structure. Authors previously studying capital structure have identified multiple types of determinants and have come up with theories explaining the differences in capital structure across firms and industries.
An important theory that authors use to explain the optimization of capital structure is the pecking order theory. As for instance explained as follows by Myers and Majluf (1984). The pecking order theory states that financing comes from three sources, namely internal funds, debt and new equity. The theory suggests that there is an order of preference when choosing between these three sources of new financing. The first option is internal financing, second is debt issuance and third is the issuance of equity. So, when internal funds are exhausted managers will need external financing, and will thereby choose employing debt over issuing new shares. The theory considers this pecking order to exist because of asymmetric information. That is, managers know more about the company than outsiders. Outsiders can only pick up the signals managers send through applying this theory. For instance, debt issuance is a positive signal and equity issuance a negative signal. Summarizing, the pecking order theory finds that companies use internal funds first, followed by debt and then equity issuance, in order to mitigate the problems associated with information asymmetry (Antonczyk and Salzmann, 2014).
5
less necessity to raise debt. Furthermore, according to the agency theory, agency cost is higher for firms with higher liquidity ratios since managers have more funds available for their personal benefit. Summarizing, a negative relationship between liquidity and a firm’s debt ratio is to be expected (Rajan and Zingales, 1995).
A second theory that is used to explain the composition of a firm’s capital structure is the trade-off theory. This theory suggests that the amount of debt financing and equity financing in a company depends on a balance between the costs and benefits. The theory assumes that there are advantages and disadvantages to financing with debt. The benefit of corporate financing with debt mainly comprises of the tax savings that can be gained through the tax deductibility of the interest payments necessary when employing debt (Modigliani and Miller, 1963). Disadvantages are the costs involved with debt financing, as for instance the required interest payments. However, the better the firm’s reputation and the longer its experience with debt financing the lower these borrowing costs are (Milton and Raviv, 1991). Also, larger firms benefit from economies of scale and have a better bargaining position than smaller firms do. Therefore, cost of debt will be lower for larger firms and they have higher levels of debt (Rajan and Zingales, 1995; Booth et al., 2001). Another factor that is found to decrease firms costs of debt is the tangibility of its assets since a firm’s tangible assets can be used as a security for the lenders. When firms have trouble repaying the debt, a firm’s assets can be used to cover for the debt and therefore can act as a security for the lenders. For that reason, when a firm has a high level of tangible assets lenders are more certain that their debt will be covered for and therefore lenders’ risk is lower. Firms that have a low level of tangible assets pay higher interest rates and firms that have high levels of tangible assets pay lower interest rates. According to the trade-off theory, the capital structure of a firm depends on the balance between the costs and benefits of a financing option. Costs are in this instance lower and for that reason it is proposed that there is a positive relationship between the tangibility of assets and the debt ratio (Rajan and Zingales, 1995; Booth et al., 2001; Gaud et al., 2005).
6
2.2 Country specific determinants
Besides the aforementioned firm specific determinants, various studies have identified other influences that are in play regarding capital structure. These influences play at a country level and are for instance the differences between legal, economic and institutional factors across countries (De Jong et al., 2008; Demirgüç-Kunt and Maksimovic, 1999; Hall, Hutchinson and Michaelas, 2004). Research into these determinants identified two groups of influences, namely formal institutional factors and informal institutional factors. The formal institutional factors include formal rules, laws, regulations and constitutions. The informal institutional factors encompass behavioral norms and culture (Gray, Kang and Yoo, 2013). Section 2.2.1 will focus on the formal factors and section 2.2.2 on the informal factors, including culture.
2.2.1 Formal factors
7
They find a negative relationship between these variables and firm leverage, as when the lenders are better protected it is less attractive for firms to borrow money.
Besides creditor protection, investor protection also plays an important role in explaining capital structure. La Porta, Lopez-De-Silanes, Shleifer and Vishny (1997) find that the development of capital markets and the level of investor protection are closely related. Their results show that countries with low investor protection have smaller and narrower capital markets and that firms in those markets use less external finance. Therefore, it is proposed that countries with low investor protection incorporated lower levels of debt. Besides investor protection the amount of rights shareholders have regarding major corporate decisions also influences capital structure. De Jong et al. (2008) find a negative relationship between shareholder rights and capital structure.
A firm’s capital structure is also influenced by a country’s legal system. La Porta et al. (1997) state that in common law countries the laws are stronger enforced and therefore influencing the legal environment. It is found that firms in common law countries have higher levels of debt than countries in non common law countries (Baker and Martin, 2011; Arosa, Richie and Schuhmann, 2014). Lastly, the economic environment and development of a country’s economy is found to have an influence on firm capital structure as well. The economic development is often proxied by authors as GDP growth and is found to have a positive influence on a firm’s debt ratio (De Jong et al., 2008; Arosa et al., 2014; Gleason et al., 2000). 2.2.2 Informal factors
In this section the informal factors that influence capital structure will be discussed. As Gray et al. (2013) suggested, these informal institutional factors encompass the behavioral norms and culture of a country’s people.
Culture is a rather broad and abstract concept and encompasses many aspects, which makes it difficult to measure. Hofstede is one of the few that has established well-accepted measures of culture in literature. He came up with multiple dimensions that measure culture, as to which countries can be compared. Of these, the most often used cultural dimensions are power distance, individualism versus collectivism, masculinity versus femininity and uncertainty avoidance. The following section explains culture based on Hofstede’s theory.
8
patterns, rituals, values, beliefs and underlying assumptions. In principle, culture is the way people think, feel and act (Hofstede, 2001). Commonly, the term culture is used for all kinds of different collectives, ranging from small groups to large groups. The term culture is most often used to describe tribes, ethnic groups, nations and organizations. Culture on some levels can be changed or easily adapted to, like for instance occupational or organizational culture. However, Hofstede (2011) describes that national cultures are much deeper rooted in the human mind since these have been acquired from an early age onwards. Societal cultures are often unconscious and are vested in values, which means that people prefer certain states of affairs over others (Hofstede, 2001).
The first dimension to measure culture, power distance, is related to the different solutions to human inequality and reflects the degree to which inequality in power distribution is expected and accepted by the less powerful in a given country. Individualism versus collectivism is related to the integration of individuals into societal groups and reflects the extent to which a society emphasizes the role of the individual versus the role of the group. The third dimension, masculinity versus femininity, is related to the emotional roles between women and men and reflects the degree to which male assertiveness is promoted as a dominant value in society versus female nurturance. And finally, uncertainty avoidance is related to the level of stress a society has when facing an unknown future or situation. It reflects the degree to which people avoid uncertain, ambiguous, unforeseeable and unstructured events (Hofstede, 1980). Based on 9 cultural dimensions, under which most of the dimensions above, the GLOBE project identified 10 cultural clusters around the world. For this project data on 61 nations’ cultural values and beliefs was used, upon which the clusters are based. The cultural clusters that were identified are Anglo Cultures, Latin Europe, Nordic Europe, Germanic Europe, Eastern Europe, Latin America, Sub-Sahara Africa, Arab Cultures, South Asia and Confucian Asia. In their work, Gupta, Hanges and Dorfman (2002) grouped countries according these cultural clusters. Countries’ cultures in a certain cluster are similar to one another and dissimilar to countries in a different cluster. Through these clusters groups of countries with similar cultures can be compared with other groups on certain characteristics or traits, as for instance Gleason et al. (2000) did. In their study they compared the mean debt ratio of four European cultural clusters and conclude that these mean ratios are significantly different from one another. However, as mentioned before data from 1994 was used and the authors controlled only for a few of the discussed firm and country specific determinants, which limits the representability.
9
investigated the effect of national culture as an aggregate on capital structure. As a means of explaining the effect of culture on capital structure the four Hofstede dimensions will be used. As discussed previously, power distance is related to the degree of inequality in a society and how it is perceived. Countries with low scores on power distance tend to place greater value on equality and interpersonal trust (Gleason et al., 2000). Zheng, El Ghoul and Guedhami (2012) find that countries with high power distance tend to issue debt with shorter maturity. This because of the suggested lower level of interpersonal trust in those countries the transaction cost of debt is higher. For that reason, it was found that countries with high scores on the cultural dimension power distance tend to rely more on market financing than on debt financing, and therefore having a negative impact on leverage (Aggarwal and Goodell, 2009; Arosa et al., 2014).
In individualistic countries, managers tend to look after their own interests more than the shareholders’ interests. Jensen and Meckling (1976) find that conflicts may arise because of the agency theory. The interests of managers and the shareholders may not always be aligned. For instance, managers might use firm resources for their own personal benefit, which obviously is not in the interest of the shareholders. Jensen (1986) calls this the free cash flow problem. These agency problems can be mitigated by debt financing (Jensen and Meckling, 1976). In less individualistic countries (i.e. collectivist countries) these agency costs are less problematic, since managers tend to look after the groups interest more than their own (Chui et al., 2002). The need for debt financing is therefore lower, indicating a positive relationship between individualism and leverage.
In masculine countries assertiveness and individual success are imperative. This leads to managers accepting projects with the highest probability of success. With increasing levels of debt the chance of bankruptcy increases as well, especially when a project goes wrong. Bankruptcy or a worsened financial position is seen as losing public image and is therefore avoided in masculine countries more than in feminine countries (Chui et al., 2002). Consequently it is proposed that in masculine countries firms employ less debt. Furthermore, an increased exposure to the risk of bankruptcy is a situation that countries which score high on uncertainty avoidance are proposed to avoid. For that reason, a negative relationship between uncertainty avoidance and leverage is expected (Gleason et al., 2000).
10
From the previous discussion the following hypotheses are derived:
H1: While controlling for other determinants, culture still has a strong effect on capital structure when comparing cultural clusters around the world.
H2: the mean leverage ratios of the European cultural clusters have converged, compared to two decades ago.
Considering the discussion of existing literature it is expected that both hypotheses will not be rejected and that significant differences among cultural clusters around the world will be found. However, it is expected that the differences among the European cultural clusters will not be as great as they were two decades ago.
3
Methodology
This section will elaborate on the methods to be used to test the hypotheses. First, the regression model will be discussed, after which a detailed description of the dependent variable and independent variables that follow from the literature will be provided. This will be followed by an elaboration on the cultural clusters and the statistical methods.
3.1 Model
The variable under study in this research is leverage and is therefore the dependent variable. As identified in the theoretical background there are various variables that determine a firm’s leverage ratio. These variables are included in the study as control variables. In this study the difference in leverage across the cultural clusters is being researched. Therefore, the previously discussed cultural clusters are included as dummy variables in the regression. The model that is used to test the hypotheses is as follows:
it j ij x xit it
LEV
CulturalCluster
Z
11 3.1.1 Dependent variable
Leverage is the dependent variable of this study and indicates a firm’s debt ratio. It is calculated by dividing the value of the firm’s liabilities by the value of the firm’s assets (Ross, Westerfield, Jaffe and Jordan, 2011). Hereby this study uses the same calculation of leverage as Gleason et al. (2000).
3.1.2 Firm level control variables
Size (+). Literature indicated size is an important factor in capital structure choices. In this study size is included as a control variable and is calculated as the logarithm of sales. Hereby the method of Rajan and Zingales (1995) and Gleason et al. (2000) is followed.
Profitability (-). Regarding profitability this study follows Rajan and Zingales (1995) method of calculating profitability by dividing the firm’s EBITDA by the book value of its assets. However, in the regression this value is lagged by one year. The reason for doing so is that it cannot be expected that managers can change their capital structure instantly, but this is merely a result of the preceding years. By lagging these values it is certain that all values are known by the firm’s managers when making capital structure decisions. Also, this might prevent problems regarding endogeneity (Getzmann, Lang and Spremann, 2010).
Tangibility (+). The variable tangibility of assets is calculated by dividing a firm’s fixed assets by the value of its total assets. Hereby the method of Gleason et al. (2000) is followed. Just as the variable profitability, the values of the variable tangibility are lagged by one period as well. Non-debt tax shield (-). This variable measures the non-debt tax shield, i.e. depreciation, firms
can employ. Following Titman and Wessels (1988), it is measured as the total value of depreciation divided by the book value of total assets. This variable is lagged by one period on its own value too.
Liquidity (-). Liquidity is measured by dividing the value of total current assets by total current liabilities. Hereby the method of Deesomsak, Paudyal and Pescetto (2004) is followed. Liquidity is also lagged by one period.
3.1.3 Country level control variables
The country data is derived from multiple different sources. In the literature a number of country specific determinants were identified. These were translated into the following variables:
12
index ranks countries on various dimensions. The dimensions investor protection and shareholder rights are two of the dimensions in this index where countries are given a score between zero and ten. These two scores are included for the two variables.
GDP growth (+). Data on countries’ GDP growth was derived from the World Bank (World Bank, 2016). This data indicates a country’s annual GDP growth in a one decimal percentage. GDP is the sum of gross value added by all producers in the country’s economy plus product taxes and minus subsidies not included in the value of the products (World Bank, 2016). The GDP growth percentage is calculated using the local currency value. The annual GDP growth is included in the dataset and is, for the same reason as with the firm variables, lagged by one period.
Creditor rights (-). The World Bank also provides scores on the strength of legal rights in a country. This index measures the degree to which collateral and bankruptcy laws protect the rights of the borrowers and lenders in a country. Countries are given scores between zero and 12. A higher score indicates that a country’s laws are better designed to expand access to credit (World Bank, 2016).
Developed country (+). Literature indicated that whether a country is a developed country or a developing country influences the capital structure of a firm in that country. A statistical annex report by the United Nations provides information on country classifications and indicates whether or not a country is a developed country (United Nations, 2012). From this report the information needed to construct the variables was derived. A dummy variable called Developed country is constructed which can only take two values; a country is either a developed country (1) or a developing country (0).
Common law country (+). Literature indicated that there are two main types of legal systems, namely common law and civil law, which both can influence capital structure. Next to these two, other types of law might be incorporated in a country’s legal system as well, as for instance Islamic law. Since existing literature only identified common law and civil law, these other types will not be incorporated in this study. Data on countries’ legal system was derived from the CIA’s World Factbook (CIA World Factbook, 2016). Since countries in this study can only take two values, either common law or civil law, this dummy variable is created. A value of 1 for this variable indicating a country is a common law country and a value of 0 indicating a country is a civil law country.
3.1.4 Clusters
13
since one needs to be left out so the remaining dummies have something to be compared to. However, specifying eight dummies automatically specifies the ninth one (Gleason et al., 2000). Following Gleason et al. (2000) the cultural cluster Nordic Europe is left out and therefore used as a benchmark for the other eight clusters. However, the results will now only indicate whether the included eight clusters are significantly different from the benchmark cluster or not. Therefore, through performing a t-test the significance of the difference in the mean leverage ratios between the other eight clusters will be tested as well. Also, a joint F-test will be performed in order to test the joint significance of the cultural clusters. Meaning that being a member of a certain cultural cluster has an effect on a firm’s leverage ratio.
3.2 Statistical method
The presented model will be tested using panel data analysis. With panel data the behavior of entities can be observed over time, with entities for instance being companies. Advantages of using this method are the ability to measure within-sample change over time and the ability to control for variables that cannot be observed or measured. These variables are for instance different business practices across companies or variables that change over time, but not across entities. Thus, it allows for individual heterogeneity. Since a lot of factors influence the financial markets and company results, it is important that year effects are controlled for. By picking one year as the year of analysis, the results might not be robust because of shocks in the financial markets and economy. Which in turn influence firm results. For that reason, panel data analysis is used, so that multiple years can be included in the analysis and year effects are controlled for. Besides being in a certain cultural cluster other variables influence firm leverage as well. Therefore, by organizing the sample as panel data the effect of these other variables can be controlled for over time. The dataset is constructed such that the entity or panel is the firm variable and the time variable is year. The dataset is unbalanced, as not all firms have complete data for all years. However, this is dealt with through only including the complete observations in the regression, thus making it balanced.
3.2.1 Fixed or random effects?
14
technique that will be used in this study. Since it is suspected that there might be correlations between the error terms within a country, the errors will be clustered per country when running the regression.
3.3 Robustness
As a robustness check a linear regression using OLS is used in order to see if the results are consistent using different statistical methods. This method does not take into account the effects over time per entity, but sees every observation as an independent observation. Just as with the panel data analysis, the errors will be clustered per country since it is suspected that there might be correlations between the error terms.
3.4 Comparison previous research
In previous literature (Gleason et al., 2000) conclusive evidence was found that capital structures vary by cultural cluster in Europe. These authors incorporated four European cultural clusters in their study. Since they base their study on data from 1994 and the European landscape has changed over the decades, this study’s results will be compared with Gleason et. al’s (2000) study. Since in their study countries were grouped into cultural clusters based on Hofstede’s dimensions and not on the work of Gupta et al. (2002) and Livermore (2013), their clusters are slightly different from this study’s clusters. For that reason, four new clusters are created that are similar to the comparison study’s clusters. In the comparison the four European clusters are now as follows. Cluster 1 will now comprise of France and Spain. Cluster 2 consists out of Ireland and Great Britain. Cluster 3 being Austria, Germany, Italy and Switzerland and cluster 4 now consists out of Denmark, Finland, and Sweden. They ran two regressions with leverage as the dependent variable. In the first regression they run the different clusters on the dependent variable leverage and in the second regression they included GDP growth, firm size and tangibility of assets as control variables. In order to compare the results, these models will be estimated again using this study’s data.
4
Data and descriptive statistics
This section elaborates on the sampling process, data and descriptive statistics. First the sampling process will be discussed, followed by outlier detection and the descriptive statistics.
4.1 Sample
15
countries country data was added later. The selection of firm data will be discussed first, followed by the selection of country data.
Following De Jong et al. (2008), this study requires firms to have at least three years of data and countries have at least ten firms to remain in the sample. The ORBIS database contains financial and company data on millions of companies. Since not all companies are representable to be selected in the sample some selection criteria were applied when selecting firms. First, firms that were of public nature, governments, states and other public authorities were excluded from the sample since these are not representable for a country’s firms. Also, following Chui et al. (2002), firms from industries that have specific leverage requirements, as for instance the financial industry, were excluded from the sample. These firms were identified based on their NACE code. The NACE code is the industry standard classification code and thus makes it possible to identify the industry a firm is in. Firms with a NACE code starting with a code between 64-68 and 92 (i.e. all financial services and gambling related companies) were excluded from the sample. Also, firms with missing data on the values needed for constructing the leverage ratio for one of the years in the period 2006-2015 were excluded from the sample. This reduced the available firms greatly. From the available firms 7,040 were randomly selected. All financial data required to compute the control variables and debt ratios were included in the data set. As previously mentioned, De Jong et al. (2008) require countries to have at least ten firms to be included in the sample. Since it is preferred to be on the safe side of that number and it is essential that a country’s average debt ratio is representative, this study requires countries to have at least 15 firms. This further decreased the sample size, since all observations from countries that had less than this number were dropped. Furthermore, even though the sample was randomly selected, it was found that almost half of the sample was made out of US based companies. This might cause the results to be biased and threatens the representability of this study. Since it is preferred that countries in the sample have equal weights, the sample size was further reduced. Through random selection countries were given equal weights, with the 39 countries now having between 16 and 24 firms and all 840 firms having 10 years of financial data available. Total number of observations is thus 8,400.
4.2 Clusters
16
are Anglo Cultures, Latin Europe, Nordic Europe, Germanic Europe, Eastern Europe, Latin America, Arab Cultures, Southern Asia and Confucian Asia. In Table 1 in the Appendix an overview of the clusters used in this study and a list of which countries belong to which cluster is provided. For each cluster a dummy variable is created, so that the differences in leverage across the nine clusters could be identified. Each observation was given a value of 1 for the cluster dummy variable it belongs to and a value of 0 for the other eight clusters dummies. Since Vietnam and Jordan are not specified in the work of Gupta et al. (2002) and Livermore (2013) these observations are excluded from the regression and further reduced the sample to 7,970 observations from 37 countries.
4.3 Outliers and missing data
Following Li, Griffin, Yue and Zhao (2011), all observations with negative values of total assets, total liabilities and sales were dropped, with the number of observations now totaling 7,956. Also following Li et al. (2011), leftover outliers were dealt with through winsorizing all firm data. Essentially, outliers are deemed as incorrect and are replaced by more likely values. All firm level variables are winsorized at the 1% level in both tails, i.e. at the 1st and 99th
percentile, of the distribution. This procedure replaces any data value above the 99th percentile
of the sample data by the value of the 99th percentile and any value below the 1st percentile by
the value of the 1st percentile. This prevents distortions in the regression results. The variables
that are winsorized are leverage, size, profitability, tangibility, non-debt tax shield and liquidity. No outliers for the country variables are to be expected, since data on the country level merely consist out of values from respectable indices or can only take two values, either 0 or 1. Observations with missing data on one of the firm variables are excluded from the regression. Not all companies have financial data for all years, resulting in the firm variables having missing data for some observations. The majority of missing data is caused by missing data on sales. Also, because of lagging these variables on their own values from the previous period there is no lagged value for the year 2006 since 2005 is not included in the sample. This further reduced the total number of observations in the sample to 5,950.
4.4 Descriptive statistics
17
their asset’s book value. The reason that not all observations were given a score on creditor rights is because Taiwan is not specified in the index the scores were derived from. Values are missing for the common law country variable, since Saudi Arabia, Kuwait and the Philippines have a legal system that cannot be characterized as either common or civil law. An overview of the values of the country variables is provided in Table 2 in the Appendix.
In Figure 4.1 the mean leverage ratios per cluster are projected over the years. At first sight one can see clear differences in the mean leverage ratios between most clusters. However, whether these are significant differences or not will be tested later. In Table 1 in the Appendix a detailed table underlying this graph is provided. In this table the mean leverage ratios and standard deviations per country are shown.
Table 4.1: Summary statistics
Variable # observations Mean
Standard
deviation Minimum Maximum
Dependent variable
Leverage 5,950 0.506 0.226 0.043 1.207
Firm level control variables
Size 5,950 12.795 2.459 -0.133 19.299
Profitability 5,950 0.100 0.130 -0.584 0.472
Tangibility 5,950 0.536 0.221 0.044 0.958
Non-debt tax shield 5,950 -0.030 0.025 -0.138 0
Liquidity 5,950 2.130 2.262 0.248 17.620
Country level control variables
GDP growth 5,950 2.720 3.503 -9.1 15.2
Creditor rights 5,869 5.368 2.617 2 12
Developed country 5,950 0.634 0.482 0 1
Common law country 5,602 0.286 0.452 0 1
Investor protection 5,950 6.421 1.048 3.8 8.3
18 Figure 4.1: Mean leverage ratios per cluster
In Table 3 in the Appendix a correlations table is provided. It shows the correlation coefficients between all variables, except the cluster dummies. It is important to check for correlations between the variables since one might suspect that there are correlations between some of the country variables that indicate legal rights and protection. A moderate positive relationship for some of the country variables was found. However, when examining the scatterplots only the relationship between investor protection and creditor rights and investor protection and shareholder rights was found to be linear. Nevertheless, no action is undertaken regarding these moderate correlations, since all three variables measure something different and should be included based on theory. Also, R2 increases by 2% per instance when adding these variables
one by one to the regression.
5
Results
In this section the estimation results will be discussed. The regression results can be seen in Table 5.1. Four regression models using random effects as the main estimation are estimated and are shown in column 1-4 respectively. In the first model only the firm level control variables are run on the dependent variable. In the models that follow additional variables are
19
being added. In the second model the country level variables are added and in the third model the cultural cluster dummies. The reason for doing so is that the strength of the model can be tested. The fourth model is run using random effects, with the errors being clustered at a country level.
5.1 Regression models
Regression (1) shows the relationship between the firm level variables and the dependent variable leverage. This model tests the explaining power of the firm level variables that are included in this study’s model as control variables. It can be seen that all firm control variables are significant a 1% level and therefore important in explaining firm’s leverage ratio. As expected a significant and positive relationship between firm size and leverage is found. Furthermore, literature indicated a negative relationship between leverage and profitability, non-debt tax shield and liquidity should exist. This negative and significant relationship was found for all these variables. However, tangibility was also found the be negative, while literature indicated a positive relationship with leverage should exist. This indicates that in this study’s sample firms with a higher level of fixed assets have lower levels of debt. The model’s R2 is 31%, indicating that the model explains 31% of the variance in the leverage.
20
captured by the common law variable. Evidently, in this study’s sample the level of a country’s legislation on being a shareholder causes leverage to be higher. The R2 is 30%, which is slightly
lower than the previous model. However, according to previous literature and theory it is expected that these variables are important in explaining firm leverage. Therefore all discussed variables stay included in the regression.
In Regression (3) the cultural cluster dummy variables are added. Adding these variables affects the influence of the developed country variable and common law country variable. The influence of being in a developed country turns from highly significant and positive to negative and only significant at the 10% level. The influence of the common law country variable turns from significant at the 10% level to significant at the 5% level. These changes might be attributed to the fact that some of the country effects are now captured by the cultural cluster dummies. For instance, all countries in the Anglo Cultures cluster are developed countries and use common law as their legal system. As mentioned in the methodology, the cultural cluster Nordic Europe is used as the benchmark cluster. Therefore, the coefficients and significance of the other eight cultural clusters indicate whether their mean leverage ratios are different from Nordic Europe’s or not. We can see that the mean leverage ratio of Nordic Europe is significantly different from the clusters Germanic Europe, Eastern Europe, Latin America, Arab Cultures, Southern Asia and Confucian Asia, but not significantly different from Anglo Cultures and Latin Europe. The cultural clusters’ coefficients that are significantly different are negative. This indicates that, ceteris paribus, the mean level of debt in those clusters is lower than in the cluster Nordic Europe. However, the real world is not a ceteris paribus setting, as other variables also have an influence. This is evidenced by the actual observed mean leverage ratios in Figure 4.1, where Nordic Europe does not have the highest leverage ratio. In fact, Latin Europe, Latin America and Germanic Europe’s mean leverage ratios are higher.
Since it is suspected that the error terms are correlated the error terms of the countries are clustered in Regression (4). Through doing so robust standard errors are obtained. The results of this regression do not differ much from Regression (3). Only the relationship between leverage and the variables size, tangibility and non-debt tax shield go from highly significant to significant at the 5% and 10% levels. R2 does not change either by clustering the
21
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 The dependent variable is firm leverage ratio.
The cultural clusters’ benchmark is the Nordic Europe cluster.
Table 5.1: Regression models (1) Random effects (2) Random effects (3) Random effects (4) Random effects
Variables Estimate (SE) Estimate (SE) Estimate (SE) (robust SE) Estimate
Size 0.0160*** 0.0137*** 0.0114*** 0.0114** (0.00189) (0.00201) (0.00204) (0.00530) Profitability -0.180*** -0.181*** -0.177*** -0.177*** (0.0177) (0.0183) (0.0183) (0.0557) Tangibility -0.0540*** -0.0464*** -0.0509*** -0.0509* (0.0150) (0.0158) (0.0157) (0.0306) Non-debt tax shield -0.298*** -0.402*** -0.375*** -0.375** (0.0954) (0.0994) (0.0994) (0.150) Liquidity -0.0198*** -0.0195*** -0.0198*** -0.0198*** (0.00104) (0.00108) (0.00108) (0.00296) Investor protection -0.0151 -0.0124 -0.0124 (0.00940) (0.0106) (0.0132) Shareholder rights 0.0140*** 0.0150*** 0.0150*** (0.00398) (0.00563) (0.00521) GDP growth 0.000314 0.000386 0.000386 (0.000556) (0.000558) (0.000968) Creditor rights -0.0212*** -0.0289*** -0.0289*** (0.00357) (0.00436) (0.00413) Developed country 0.0607*** -0.0736* -0.0736* (0.0175) (0.0392) (0.0417)
Common law country 0.0371* 0.0643** 0.0643**
22
5.2 Differences in cultural clusters
This study’s purpose is to find whether or not culture influences firm leverage by proposing that firm leverage varies by cultural cluster. From the model previously discussed only the differences between the benchmark cluster Nordic Europe and the other eight clusters can be derived. From this model the differences among the other cultural clusters cannot be derived. For that reason a t-test and joint F-test will be run. The F-test tests whether or not the cultural clusters’ coefficients together are equal to zero. The test indicates that together they are highly significant, which indicates they jointly have a significant effect on leverage.
By performing the t-test it is checked whether the cultural clusters are significantly different from one another. As said, the significance of the differences between the cluster Nordic Europe and the other eight clusters is shown in Table 5.1: Regression (4). The results of the t-test indicate the differences between the remaining cultural clusters, which are shown in Table 5.2.
Table 5.2: Significance differences mean leverage clusters
Cultural clusters 1 2 3 4 5 6 7 8 9 1. Nordic Europe - 2. Anglo Cultures - - 3. Latin Europe - - - 4. Germanic Europe ** - *** - 5. Eastern Europe ** - *** - - 6. Latin America *** *** *** ** ** - 7. Arab Cultures *** *** *** *** *** - - 8. Southern Asia *** *** *** *** *** - - - 9. Confucian Asia *** *** *** ** - - *** * - *** p<0.01, ** p<0.05, * p<0.1
In this table the results of the t-test are presented, indicating whether the mean leverage ratio of the given cluster differs from the other eight clusters. For instance, one can see that cluster 2 (Anglo Cultures) is significantly different from cultural clusters Latin America, Arab Cultures, Southern Asia and Confucian Asia at the 1% level, but not significantly different from the other clusters. Also, it is found that cultural cluster Southern Asia is only significantly different from Confucian Asia at the 10% level.
5.3 Robustness
23
variables are also included. In the third model the cultural clusters are added and in the fourth model the errors are clustered per country. In the OLS models year effects are not accounted for.
These models show similar results as the previous models. Besides some slight changes in the height of the coefficients and significance of some variables when adding extra variables in models 2 and 3, no drastic differences with the random effects models were found. When comparing both clustered models in the random effects and OLS setting, only the variables non-debt tax shield and developed country were found to become insignificant when denying time variant influences with the OLS model. However, most important, when testing the differences in mean leverage ratio among the cultural clusters no major differences were found with the random effects model and results seem to be robust.
However, when using the same four European cultural clusters and the same variables as Gleason et al. (2000) used, different results are found. The estimation results are presented in Table 5.3. Contrary to Gleason et al. (2000) no such conclusive evidence that the four European cultural clusters vary was found. As shown in Table 5.3, the results only show a significant difference between Cluster 1 and the benchmark cluster, Cluster 4. After testing the relationships between the other clusters only the difference between the mean leverage ratio of cluster 1 and 4 were found to be significant. Also, they find that that the average leverage ratio of cluster 4 is 33.48% and cluster 2 18.19%. While in this present study these percentages are 54% and 56%, respectively. However, note that the previous study used a sample of retail companies and this study uses retail plus companies from other industries. As introduced previously, these differences in results might be attributed to the fact that the European landscape changed a lot over the past decades, seemingly resulting in culture having a smaller impact in Europe than before.
Table 5.3: Results comparison European clusters
(1) Random effects (2) Random effects Variables (robust SE) Estimate (robust SE) Estimate
Cluster 1 0.135*** 0.0888*** (0.0301) (0.0289) Cluster 2 0.0242 0.0264 (0.0188) (0.0225) Cluster 3 0.0341 0.0220 (0.0464) (0.0440) GDP growth 0.00144** (0.000711) Size 0.0302*** (0.00981) Tangibility 0.0997*** (0.0328) Constant 0.536*** 0.0766 (0.0187) (0.128) R2 0.07 0.22 Observations 2,009 1,995 Number of companies 224 224
24
6
Conclusion
The purpose of this study was to find to what extent culture influences capital structure decisions around the world. In addition, this study sought to find whether culture is still influencing capital structure in Europe as intensively as two decades ago.
Literature has shown that there are many factors influencing managerial decision making regarding capital structure and the level of debt in firms. The determinants of capital structure that were found in previous literature could be categorized as either firm level determinants or country level determinants. These established determinants of capital structure were included as control variables in this study, with firm level control variables for instance being profitability or size and country level determinants the a country’s level of creditor protection or its legal system. Cultural dimensions were also found to be a determinant of capital structure, but on an aggregate level cultures have not often been compared on their capital structure.
25
Finland and Sweden. All the other differences in mean leverage ratios are found to be insignificant, while a previous study indicated that the European clusters were significantly different. For that reason, based on this study’s results, the relationship between capital structure and culture in Europe appears not to be as intensive anymore compared to two decades ago. This might be attributed to the fact that the EU has gained influence in Europe and the Economic and Monetary Union implemented the Euro since then. These results are in line with Bruno and Shin’s (2014) findings that financial conditions and risk taking across sectors and regions are synchronizing. Evidently, Europe is one of those regions.
This study’s contribution to existing literature is captured in the fact that not only differences in capital structure across European cultural clusters are being reviewed, but cultural clusters around the world are being compared. This is particularly interesting since the existing literature on culture and capital structure tend to only examine the effect of single cultural dimensions on capital structure and generally neglect the effect of cultural clusters. To the best of my knowledge, the relationship between capital structure and cultural clusters around the world has not been studied before. Through studying the effect of clusters around the world, more diversity in cultures is captured and results are now more representative. Furthermore, as indicated by Gianetti (2003), many studies on capital structure neglected to include smaller companies and literature is often based on large multinational companies. For that reason, this study included both small and large firms from multiple industries. Which is important since this makes results robust to an industry effect and not only valid for one industry. Furthermore, this study found that culture became of lesser influence in explaining differences in capital structure in Europe. This is interesting since literature is still based on research that found an important effect of cultural dimensions, while as evidenced in this study, this effect is diminishing. These results’ implication for managers is that different countries or groups of countries have different standards regarding capital structure. In the case of a merger or acquisition this means that this is something that needs to be taken into account.
26
27
7 References
Aggarwal, R., Goodell, J., 2009. Markets versus institutions in developing countries: national attributes as determinants. Emerging Markets Review 10, 51-66.
Antonczyk, R., Salzmann, A., 2014. Overconfidence and optimism: the effect of national culture on capital structure. Research in International Business and Finance 31, 132–151. Arosa, C., Richie, N., Schuhmann, P., 2014. The impact of culture on market timing in capital
structure choices. Research in International Business and Finance 31, 178–192.
Baker, H., Martin, G., 2011. Capital Structure and Corporate Financing Decisions: Theory, Evidence, and Practice. John Wiley & Sons, New Jersey.
Barton, S.L., Gordon, P.J., 1988. Corporate strategy and capital structure. Strategic Management Journal 9, 623–632.
Booth, L., Aivazian, V., Demirguc-kunt, A., Maksimovic, V., 2001. Capital structures in developing countries. Journal of Finance 56, 87–130.
Bruno, V., Shin, H., 2014. Globalization of corporate risk taking. Journal of International Business Studies 45, 800–820.
Bureau van Dijk, 2016. ORBIS: company information across the globe. Retrieved from http://orbis.bvdinfo.com/
CIA, 2016. The world factbook. Retrieved from https://www.cia.gov/library/publications/the- world-factbook/
Chui, A., Lloyd, A., Kwok, C., 2002. The determination of capital structure: is national culture a missing piece to the puzzle? Journal of International Business Studies 33, 99–127. de Jong, A., Kabir, R., Nguyen, T., 2008. Capital structure around the world: The roles of firm-
and country-specific determinants. Journal of Banking and Finance 32, 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, 387–405.
Demirgüç-Kunt, A., Maksimovic, V., 1999. Institutions, financial markets, and firm debt maturity. Journal of Financial Economics 54, 295–336.
28
Getzmann, A., Lang, S., Spremann, K., 2010. Determinants of the target capital structure and adjustment speed – evidence from Asian capital markets. Unpublished working paper. University of St. Gallen, Gallen.
Giannetti, M., 2003. Do better institutions mitigate agency problems? Evidence from corporate finance choices. Journal of Financial and Quantative Analysis 38, 185-212 Gleason, K., Mathur, L., Mathur, I., 2000. The interrelationship between culture, capital
structure, and performance: evidence from European retailers. Journal of Business Research 50, 185–191.
Gray, S., Kang, T., Yoo, Y., 2013. National culture and international differences in the cost of equity capital. Management International Review 53, 899–916.
Gupta, V., Hanges, P., Dorfman, P., 2002. Cultural clusters: methodology and findings. Journal of World Business 37, 11–15.
Hall, G., Hutchinson, P., Michaelas, N., 2004. Determinants of the capital structures of European SMEs. Journal of Business Finance and Accounting 31, 711–728.
Hirshleifer, D., Thakor, A., 1992. Managerial conservatism, project choice, and debt. The Review of Financial Studies 5, 437–470.
Hofstede, G., 1980. Culture's Consequences: International Differences in Work-Related Values. Sage Publications Inc., California
Hofstede, G., 2001. Culture’s Consequences: Comparing Values, Behaviors, Institutions, and Organizations Across Nations. Sage Publications Inc., California
Hofstede, G., 2011. Dimensionalizing cultures: the Hofstede model in context. Online Readings in Psychology and Culture 2, 1–26.
Jensen, M., 1986. Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review 76, 323–329.
Jensen, M., Meckling, W., 1976. Theory of the firm: managerial behavior , agency costs and ownershipstructure. Journal of Financial Economics 3, 305–360.
La Porta, R., Lopez-de-silanes, F., Shleifer, A., Vishny, R., 1997. Legal determinants of external finance. Journal of Finance 52, 1131–1151.
29
Livermore, D., 2013. Expand Your Borders: Discover Ten Cultural Clusters. Cultural Intelligence Center LLC, Michigan.
Milton, H., Raviv, A., 1991. The Theory of Capital Structure. Journal of Finance 46, 297–355. Modigliani, F., Miller, M., 1963. Corporate income taxes and the cost of capital: a correction.
American Economic Review 53, 433–443.
Modigliani, F., Miller, M., 1958. The cost of capital, corporation finance and the theory of investment. American Economic Review 48, 261–297.
Myers, S., Majluf, N., 1984. Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics 13, 187–221. Rajan, R., Zingales, L., 1995. What do we know about capital structure? Some evidence from
international Data. Journal of Finance 50, 1421–1460.
Ross, S., Westerfield, R., Jaffe, J., Jordan, B., 2011. Corporate Finance: Core Principles and Applications. McGraw-Hill Education, New York.
Song, J., Philippatos, G., 2004. Have we resolved some critical issues related to international capital structure? Empirical evidence from the 30 OECD countries. Unpublished working
paper, University of Tennessee, Knoxville.
Titman, S., Wessels, R., 1988. The determinants of capital structure choice. Journal of Finance 43, 1 – 19.
United Nations, 2012. Statistical Annex. Retrieved from http://www.un.org/en/development /desa/policy/wesp/wesp_current/2012country_class.pdf. Accessed on 15-03-2016.
Wang, D., Esqueda, O., 2014. National cultural effects on leverage decisions: evidence from emerging-market ADRs. Research in International Business and Finance 31, 152–177. World Bank, 2016. GDP growth. Retrieved from http://data.worldbank.org/indicator/NY. GDP.MKTP.KD.ZG. Accessed on 15-03-2016.
World Bank, 2016. Strength of legal rights index. Retrieved from http://data.worldbank.org/ indicator/IC.LGL.CRED.XQ. Accessed on 15-03-2016.
World Bank Group, 2016. Doing Business: measuring business regulations. Retrieved from http://www.doingbusiness.org/data/exploretopics/protecting-minority-investors. Accessed on 15-03-2016.
30
Appendix
Table 1: Mean leverage ratio and descriptive statistics per country
Cluster Country Mean St. Dev. [95% Conf. Interval] # obs.
Anglo Cultures Australia 0.310 0.017 0.276 0.345 230
Canada 0.436 0.017 0.403 0.470 220 Great Britain 0.558 0.017 0.524 0.591 210 Ireland 0.565 0.016 0.533 0.597 200 New Zealand 0.434 0.010 0.413 0.454 230 United States 0.592 0.016 0.560 0.624 220 South Africa 0.459 0.012 0.435 0.484 210
Latin Europe Spain 0.703 0.012 0.678 0.727 220
France 0.644 0.012 0.621 0.668 230
Israel 0.535 0.017 0.501 0.570 230
Italy 0.691 0.012 0.667 0.714 210
Nordic Europe Denmark 0.537 0.012 0.513 0.561 240
Finland 0.565 0.011 0.543 0.588 208
Sweden 0.477 0.015 0.446 0.508 230
Germanic Europe Austria 0.577 0.011 0.554 0.599 160
Switzerland 0.474 0.010 0.453 0.496 230
Germany 0.554 0.010 0.534 0.575 220
Eastern Europe Greece 0.575 0.017 0.542 0.609 170
Poland 0.470 0.013 0.444 0.496 230
Russia 0.566 0.020 0.525 0.607 180
Latin America Brazil 0.589 0.010 0.568 0.610 230
Chile 0.544 0.008 0.526 0.561 200
Arab Cultures United Arab
Emirates 0.321 0.016 0.289 0.352 200
Egypt 0.499 0.015 0.469 0.530 258
Kuwait 0.395 0.016 0.362 0.428 160
Saudi Arabia 0.376 0.013 0.349 0.402 230
Southern Asia Indonesia 0.477 0.012 0.452 0.502 230
Malaysia 0.331 0.011 0.308 0.354 240
Philippines 0.472 0.018 0.436 0.509 170
Pakistan 0.607 0.017 0.573 0.642 220
Thailand 0.494 0.018 0.459 0.530 230
Confucian Asia China 0.494 0.012 0.470 0.519 230
31
The scores on investor protection and shareholder rights are on a scale of 0-10 and scores on creditor rights from 0-12, with 10 and 12 indicating high levels of protection/rights. GDP growth shows the average annual growth percentage for the years 2006-2015. Table 2: Values country variables
Cluster Country Investor
protection Shareholder rights Creditor rights Developed country Common law GDP growth
Anglo Cultures Australia 5.7 5 11 Yes Yes 2.80
Canada 7.7 6 9 Yes Yes 1.76
Great Britain 7.8 8 7 Yes Yes 1.16
Ireland 7.3 7 7 Yes Yes 1.53
New Zealand 8.3 8 12 Yes Yes 1.69
United States 6.5 4 11 Yes Yes 1.38
South Africa 7.2 8 5 No Yes 2.76
Latin Europe Spain 6.5 10 5 Yes No 0.18
France 6.5 6 4 Yes No 0.81
Israel 7.3 7 6 Yes No 3.96
Italy 6.3 8 2 Yes No -0.62
Nordic Europe Denmark 6.8 8 8 Yes No 0.17
Finland 5.7 7 7 Yes No 0.49
Sweden 7.2 9 6 Yes No 1.58
Germanic Europe Austria 6.3 8 5 Yes No 1.21
Switzerland 5 8 6 Yes No 1.99
Germany 6 8 6 Yes No 1.40
Eastern Europe Greece 6.2 8 3 Yes No -2.22
Poland 6 8 7 Yes No 3.87
Russia 5.7 7 6 No No 3.13
Latin America Brazil 6.5 7 2 No No 3.50
Chile 6.3 10 4 No No 3.90
Arab Cultures United Arab Emirates 6 4 2 No No 3.73
Egypt 4.5 2 2 No No 4.36
Southern Asia Indonesia 5.3 7 5 No No 5.71
Malaysia 7.8 6 7 No Yes 4.90
Pakistan 6.7 8 3 No Yes 3.60
Thailand 6.3 5 3 No No 3.41
Confucian Asia China 4.3 1 4 No No 9.84
Hong Kong 8.3 9 8 Yes Yes 3.56
Japan 6.3 8 4 Yes No 0.54
South Korea 7.3 7 5 Yes No 3.66
32 Table 3: Correlation table
Variable 1 2 3 4 5 6 7 8 9 10 11 12
1. Leverage 1.0000
2. Size 0.3355 1.0000
3. Profitability -0.0185 0.3540 1.0000
4. Tangibility -0.0230 0.1647 0.0375 1.0000
5. Non-debt tax shield -0.0737 -0.1286 -0.1308 -0.2078 1.0000
33 Table 4: OLS regression models
(1)
OLS OLS (2) OLS (3) OLS (4)
Variables Estimate (SE) Estimate (SE) Estimate (SE) (robust SE) Estimate
Size 0.0239*** 0.0198*** 0.0160*** 0.0160*** (0.0011) (0.0012) (0.0012) (0.0047) Profitability -0.2920*** -0.2650*** -0.2380*** -0.2380*** (0.0197) (0.0203) (0.0202) (0.0760) Tangibility -0.1800*** -0.1650*** -0.1680*** -0.1680*** (0.0114) (0.0119) (0.0118) (0.0429) Non-debt tax shield -0.1200 -0.2820*** -0.1450 -0.1450
(0.0979) (0.1010) (0.1020) (0.2050) Liquidity -0.0458*** -0.0440*** -0.0439*** -0.0439*** (0.0012) (0.0012) (0.0012) (0.0071) Investor protection -0.0104*** -0.0083* -0.0083 (0.0039) (0.0044) (0.0096) Shareholder rights 0.0090*** 0.0090*** 0.0090** (0.0017) (0.0024) (0.0041) GDP growth -0.0022*** -0.0008 -0.0008 (0.0008) (0.0009) (0.0012) Creditor rights -0.0170*** -0.0238*** -0.0238*** (0.0015) (0.0019) (0.0033) Developed/developing 0.0415*** -0.0454*** -0.0454 (0.0077) (0.0166) (0.0325) Common/civil law 0.0498*** 0.0612*** 0.0612*** (0.0090) (0.0124) (0.0211) Anglo Cultures -0.0079 -0.0079 (0.0148) (0.0221) Latin Europe 0.0148 0.0148 (0.0118) (0.0236) Germanic Europe -0.0555*** -0.0555** (0.0119) (0.0265) Eastern Europe -0.0617*** -0.0617** (0.0138) (0.0285) Latin America -0.1320*** -0.1320*** (0.0224) (0.0454) Arab Cultures -0.1970*** -0.1970*** (0.0199) (0.0543) Southern Asia -0.1350*** -0.1350*** (0.0205) (0.0398) Confucian Asia -0.1080*** -0.1080*** (0.0125) (0.0240) Constant 0.7150*** 0.7150*** (0.0311) (0.0903) R2 0.33 0.35 0.38 0.38 Observations 5,950 5,521 5,521 5,521
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
The dependent variable is firm leverage ratio.
